@promptbook/browser 0.103.0-46 → 0.103.0-48

package/umd/index.umd.js

@@ -1,12 +1,13 @@
 (function (global, factory) {
-    typeof exports === 'object' && typeof module !== 'undefined' ? factory(exports, require('spacetrim'), require('crypto'), require('path'), require('destroyable')) :
-    typeof define === 'function' && define.amd ? define(['exports', 'spacetrim', 'crypto', 'path', 'destroyable'], factory) :
-    (global = typeof globalThis !== 'undefined' ? globalThis : global || self, factory(global["promptbook-browser"] = {}, global.spaceTrim, global.crypto, global.path, global.destroyable));
-})(this, (function (exports, spaceTrim, crypto, path, destroyable) { 'use strict';
+    typeof exports === 'object' && typeof module !== 'undefined' ? factory(exports, require('spacetrim'), require('crypto'), require('crypto-js'), require('crypto-js/enc-hex'), require('path'), require('destroyable')) :
+    typeof define === 'function' && define.amd ? define(['exports', 'spacetrim', 'crypto', 'crypto-js', 'crypto-js/enc-hex', 'path', 'destroyable'], factory) :
+    (global = typeof globalThis !== 'undefined' ? globalThis : global || self, factory(global["promptbook-browser"] = {}, global.spaceTrim, global.crypto, global.cryptoJs, global.hexEncoder, global.path, global.destroyable));
+})(this, (function (exports, spaceTrim, crypto, cryptoJs, hexEncoder, path, destroyable) { 'use strict';
 
     function _interopDefaultLegacy (e) { return e && typeof e === 'object' && 'default' in e ? e : { 'default': e }; }
 
     var spaceTrim__default = /*#__PURE__*/_interopDefaultLegacy(spaceTrim);
+    var hexEncoder__default = /*#__PURE__*/_interopDefaultLegacy(hexEncoder);
 
     // ⚠️ WARNING: This code has been generated so that any manual changes will be overwritten
     /**
@@ -22,7 +23,7 @@
      * @generated
      * @see https://github.com/webgptorg/promptbook
      */
-    const PROMPTBOOK_ENGINE_VERSION = '0.103.0-46';
+    const PROMPTBOOK_ENGINE_VERSION = '0.103.0-48';
     /**
      * TODO: string_promptbook_version should be constrained to the all versions of Promptbook engine
      * Note: [💞] Ignore a discrepancy between file name and entity name
@@ -383,9 +384,6 @@
             throw new Error(`${channelName} channel is greater than 255, it is ${value}`);
         }
     }
-    /**
-     * TODO: [🧠][🚓] Is/which combination it better to use asserts/check, validate or is utility function?
-     */
 
     /**
      * Color object represents an RGB color with alpha channel
@@ -1908,6 +1906,8 @@
     /**
      * Normalizes a given text to camelCase format.
      *
+     * Note: [🔂] This function is idempotent.
+     *
      * @param text The text to be normalized.
      * @param _isFirstLetterCapital Whether the first letter should be capitalized.
      * @returns The camelCase formatted string.
@@ -1996,447 +1996,690 @@
      */
 
     /**
-     * Generates a regex pattern to match a specific commitment
-     *
-     * Note: It always creates new Regex object
-     * Note: Uses word boundaries to ensure only full words are matched (e.g., "PERSONA" matches but "PERSONALITY" does not)
-     *
-     * @private - TODO: [🧠] Maybe should be public?
-     */
-    function createCommitmentRegex(commitment) {
-        const escapedCommitment = commitment.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
-        const keywordPattern = escapedCommitment.split(/\s+/).join('\\s+');
-        const regex = new RegExp(`^\\s*(?<type>${keywordPattern})\\b\\s+(?<contents>.+)$`, 'gim');
-        return regex;
-    }
-    /**
-     * Generates a regex pattern to match a specific commitment type
-     *
-     * Note: It just matches the type part of the commitment
-     * Note: It always creates new Regex object
-     * Note: Uses word boundaries to ensure only full words are matched (e.g., "PERSONA" matches but "PERSONALITY" does not)
+     * Computes SHA-256 hash of the agent source
      *
-     * @private
+     * @public exported from `@promptbook/core`
      */
-    function createCommitmentTypeRegex(commitment) {
-        const escapedCommitment = commitment.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
-        const keywordPattern = escapedCommitment.split(/\s+/).join('\\s+');
-        const regex = new RegExp(`^\\s*(?<type>${keywordPattern})\\b`, 'gim');
-        return regex;
+    function computeAgentHash(agentSource) {
+        return cryptoJs.SHA256(hexEncoder__default["default"].parse(agentSource /* <- TODO: !!!!! spaceTrim */)).toString( /* hex */);
     }
 
     /**
-     * Base implementation of CommitmentDefinition that provides common functionality
-     * Most commitments can extend this class and only override the applyToAgentModelRequirements method
+     * Tests if given string is valid file path.
      *
-     * @private
+     * Note: This does not check if the file exists only if the path is valid
+     * @public exported from `@promptbook/utils`
      */
-    class BaseCommitmentDefinition {
-        constructor(type) {
-            this.type = type;
-        }
-        /**
-         * Creates a regex pattern to match this commitment in agent source
-         * Uses the existing createCommitmentRegex function as internal helper
-         */
-        createRegex() {
-            return createCommitmentRegex(this.type);
-        }
-        /**
-         * Creates a regex pattern to match just the commitment type
-         * Uses the existing createCommitmentTypeRegex function as internal helper
-         */
-        createTypeRegex() {
-            return createCommitmentTypeRegex(this.type);
+    function isValidFilePath(filename) {
+        if (typeof filename !== 'string') {
+            return false;
         }
-        /**
-         * Helper method to create a new requirements object with updated system message
-         * This is commonly used by many commitments
-         */
-        updateSystemMessage(requirements, messageUpdate) {
-            const newMessage = typeof messageUpdate === 'string' ? messageUpdate : messageUpdate(requirements.systemMessage);
-            return {
-                ...requirements,
-                systemMessage: newMessage,
-            };
+        if (filename.split('\n').length > 1) {
+            return false;
         }
-        /**
-         * Helper method to append content to the system message
-         */
-        appendToSystemMessage(requirements, content, separator = '\n\n') {
-            return this.updateSystemMessage(requirements, (currentMessage) => {
-                if (!currentMessage.trim()) {
-                    return content;
-                }
-                return currentMessage + separator + content;
-            });
+        // Normalize slashes early so heuristics can detect path-like inputs
+        const filenameSlashes = filename.replace(/\\/g, '/');
+        // Reject strings that look like sentences (informational text)
+        // Heuristic: contains multiple spaces and ends with a period, or contains typical sentence punctuation
+        // But skip this heuristic if the string looks like a path (contains '/' or starts with a drive letter)
+        if (filename.trim().length > 60 && // long enough to be a sentence
+            /[.!?]/.test(filename) && // contains sentence punctuation
+            filename.split(' ').length > 8 && // has many words
+            !/\/|^[A-Z]:/i.test(filenameSlashes) // do NOT treat as sentence if looks like a path
+        ) {
+            return false;
         }
-        /**
-         * Helper method to add a comment section to the system message
-         * Comments are lines starting with # that will be removed from the final system message
-         * but can be useful for organizing and structuring the message during processing
-         */
-        addCommentSection(requirements, commentTitle, content, position = 'end') {
-            const commentSection = `# ${commentTitle.toUpperCase()}\n${content}`;
-            if (position === 'beginning') {
-                return this.updateSystemMessage(requirements, (currentMessage) => {
-                    if (!currentMessage.trim()) {
-                        return commentSection;
-                    }
-                    return commentSection + '\n\n' + currentMessage;
-                });
-            }
-            else {
-                return this.appendToSystemMessage(requirements, commentSection);
-            }
+        // Absolute Unix path: /hello.txt
+        if (/^(\/)/i.test(filenameSlashes)) {
+            // console.log(filename, 'Absolute Unix path: /hello.txt');
+            return true;
         }
-    }
-
-    /**
-     * ACTION commitment definition
-     *
-     * The ACTION commitment defines specific actions or capabilities that the agent can perform.
-     * This helps define what the agent is capable of doing and how it should approach tasks.
-     *
-     * Example usage in agent source:
-     *
-     * ```book
-     * ACTION Can generate code snippets and explain programming concepts
-     * ACTION Able to analyze data and provide insights
-     * ```
-     *
-     * @private [🪔] Maybe export the commitments through some package
-     */
-    class ActionCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'ACTION') {
-            super(type);
+        // Absolute Windows path: C:/ or C:\ (allow spaces and multiple dots in filename)
+        if (/^[A-Z]:\/.+$/i.test(filenameSlashes)) {
+            // console.log(filename, 'Absolute Windows path: /hello.txt');
+            return true;
         }
-        /**
-         * Short one-line description of ACTION.
-         */
-        get description() {
-            return 'Define agent capabilities and actions it can perform.';
+        // Relative path: ./hello.txt
+        if (/^(\.\.?\/)+/i.test(filenameSlashes)) {
+            // console.log(filename, 'Relative path: ./hello.txt');
+            return true;
         }
-        /**
-         * Markdown documentation for ACTION commitment.
-         */
-        get documentation() {
-            return spaceTrim.spaceTrim(`
-            # ${this.type}
-
-            Defines specific actions or capabilities that the agent can perform.
-
-            ## Key aspects
-
-            - Both terms work identically and can be used interchangeably.
-            - Each action adds to the agent's capability list.
-            - Actions help users understand what the agent can do.
-
-            ## Examples
-
-            \`\`\`book
-            Code Assistant
-
-            PERSONA You are a programming assistant
-            ACTION Can generate code snippets and explain programming concepts
-            ACTION Able to debug existing code and suggest improvements
-            ACTION Can create unit tests for functions
-            \`\`\`
-
-            \`\`\`book
-            Data Scientist
-
-            PERSONA You are a data analysis expert
-            ACTION Able to analyze data and provide insights
-            ACTION Can create visualizations and charts
-            ACTION Capable of statistical analysis and modeling
-            KNOWLEDGE Data analysis best practices and statistical methods
-            \`\`\`
-        `);
+        // Allow paths like foo/hello
+        if (/^[^/]+\/[^/]+/i.test(filenameSlashes)) {
+            // console.log(filename, 'Allow paths like foo/hello');
+            return true;
         }
-        applyToAgentModelRequirements(requirements, content) {
-            const trimmedContent = content.trim();
-            if (!trimmedContent) {
-                return requirements;
-            }
-            // Add action capability to the system message
-            const actionSection = `Capability: ${trimmedContent}`;
-            return this.appendToSystemMessage(requirements, actionSection, '\n\n');
+        // Allow paths like hello.book
+        if (/^[^/]+\.[^/]+$/i.test(filenameSlashes)) {
+            // console.log(filename, 'Allow paths like hello.book');
+            return true;
         }
+        return false;
     }
     /**
-     * Note: [💞] Ignore a discrepancy between file name and entity name
+     * TODO: [🍏] Implement for MacOs
      */
 
     /**
-     * DELETE commitment definition
-     *
-     * The DELETE commitment (and its aliases CANCEL, DISCARD, REMOVE) is used to
-     * remove or disregard certain information or context. This can be useful for
-     * overriding previous commitments or removing unwanted behaviors.
-     *
-     * Example usage in agent source:
+     * Tests if given string is valid URL.
      *
-     * ```book
-     * DELETE Previous formatting requirements
-     * CANCEL All emotional responses
-     * DISCARD Technical jargon explanations
-     * REMOVE Casual conversational style
-     * ```
+     * Note: [🔂] This function is idempotent.
+     * Note: Dataurl are considered perfectly valid.
+     * Note: There are two similar functions:
+     * - `isValidUrl` which tests any URL
+     * - `isValidPipelineUrl` *(this one)* which tests just promptbook URL
      *
-     * @private [🪔] Maybe export the commitments through some package
+     * @public exported from `@promptbook/utils`
      */
-    class DeleteCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type) {
-            super(type);
+    function isValidUrl(url) {
+        if (typeof url !== 'string') {
+            return false;
         }
-        /**
-         * Short one-line description of DELETE/CANCEL/DISCARD/REMOVE.
-         */
-        get description() {
-            return 'Remove or **disregard** certain information, context, or previous commitments.';
+        try {
+            if (url.startsWith('blob:')) {
+                url = url.replace(/^blob:/, '');
+            }
+            const urlObject = new URL(url /* because fail is handled */);
+            if (!['http:', 'https:', 'data:'].includes(urlObject.protocol)) {
+                return false;
+            }
+            return true;
         }
-        /**
-         * Markdown documentation for DELETE commitment.
-         */
-        get documentation() {
-            return spaceTrim.spaceTrim(`
-            # DELETE (CANCEL, DISCARD, REMOVE)
-
-            A commitment to remove or disregard certain information or context. This can be useful for overriding previous commitments or removing unwanted behaviors.
-
-            ## Aliases
+        catch (error) {
+            return false;
+        }
+    }
 
-            - \`DELETE\` - Remove or eliminate something
-            - \`CANCEL\` - Cancel or nullify something
-            - \`DISCARD\` - Discard or ignore something
-            - \`REMOVE\` - Remove or take away something
-
-            ## Key aspects
-
-            - Multiple delete commitments can be used to remove different aspects.
-            - Useful for overriding previous commitments in the same agent definition.
-            - Can be used to remove inherited behaviors from base personas.
-            - Helps fine-tune agent behavior by explicitly removing unwanted elements.
+    const defaultDiacriticsRemovalMap = [
+        {
+            base: 'A',
+            letters: '\u0041\u24B6\uFF21\u00C0\u00C1\u00C2\u1EA6\u1EA4\u1EAA\u1EA8\u00C3\u0100\u0102\u1EB0\u1EAE\u1EB4\u1EB2\u0226\u01E0\u00C4\u01DE\u1EA2\u00C5\u01FA\u01CD\u0200\u0202\u1EA0\u1EAC\u1EB6\u1E00\u0104\u023A\u2C6F',
+        },
+        { base: 'AA', letters: '\uA732' },
+        { base: 'AE', letters: '\u00C6\u01FC\u01E2' },
+        { base: 'AO', letters: '\uA734' },
+        { base: 'AU', letters: '\uA736' },
+        { base: 'AV', letters: '\uA738\uA73A' },
+        { base: 'AY', letters: '\uA73C' },
+        {
+            base: 'B',
+            letters: '\u0042\u24B7\uFF22\u1E02\u1E04\u1E06\u0243\u0182\u0181',
+        },
+        {
+            base: 'C',
+            letters: '\u0043\u24B8\uFF23\u0106\u0108\u010A\u010C\u00C7\u1E08\u0187\u023B\uA73E',
+        },
+        {
+            base: 'D',
+            letters: '\u0044\u24B9\uFF24\u1E0A\u010E\u1E0C\u1E10\u1E12\u1E0E\u0110\u018B\u018A\u0189\uA779\u00D0',
+        },
+        { base: 'DZ', letters: '\u01F1\u01C4' },
+        { base: 'Dz', letters: '\u01F2\u01C5' },
+        {
+            base: 'E',
+            letters: '\u0045\u24BA\uFF25\u00C8\u00C9\u00CA\u1EC0\u1EBE\u1EC4\u1EC2\u1EBC\u0112\u1E14\u1E16\u0114\u0116\u00CB\u1EBA\u011A\u0204\u0206\u1EB8\u1EC6\u0228\u1E1C\u0118\u1E18\u1E1A\u0190\u018E',
+        },
+        { base: 'F', letters: '\u0046\u24BB\uFF26\u1E1E\u0191\uA77B' },
+        {
+            base: 'G',
+            letters: '\u0047\u24BC\uFF27\u01F4\u011C\u1E20\u011E\u0120\u01E6\u0122\u01E4\u0193\uA7A0\uA77D\uA77E',
+        },
+        {
+            base: 'H',
+            letters: '\u0048\u24BD\uFF28\u0124\u1E22\u1E26\u021E\u1E24\u1E28\u1E2A\u0126\u2C67\u2C75\uA78D',
+        },
+        {
+            base: 'I',
+            letters: '\u0049\u24BE\uFF29\u00CC\u00CD\u00CE\u0128\u012A\u012C\u0130\u00CF\u1E2E\u1EC8\u01CF\u0208\u020A\u1ECA\u012E\u1E2C\u0197',
+        },
+        { base: 'J', letters: '\u004A\u24BF\uFF2A\u0134\u0248' },
+        {
+            base: 'K',
+            letters: '\u004B\u24C0\uFF2B\u1E30\u01E8\u1E32\u0136\u1E34\u0198\u2C69\uA740\uA742\uA744\uA7A2',
+        },
+        {
+            base: 'L',
+            letters: '\u004C\u24C1\uFF2C\u013F\u0139\u013D\u1E36\u1E38\u013B\u1E3C\u1E3A\u0141\u023D\u2C62\u2C60\uA748\uA746\uA780',
+        },
+        { base: 'LJ', letters: '\u01C7' },
+        { base: 'Lj', letters: '\u01C8' },
+        { base: 'M', letters: '\u004D\u24C2\uFF2D\u1E3E\u1E40\u1E42\u2C6E\u019C' },
+        {
+            base: 'N',
+            letters: '\u004E\u24C3\uFF2E\u01F8\u0143\u00D1\u1E44\u0147\u1E46\u0145\u1E4A\u1E48\u0220\u019D\uA790\uA7A4',
+        },
+        { base: 'NJ', letters: '\u01CA' },
+        { base: 'Nj', letters: '\u01CB' },
+        {
+            base: 'O',
+            letters: '\u004F\u24C4\uFF2F\u00D2\u00D3\u00D4\u1ED2\u1ED0\u1ED6\u1ED4\u00D5\u1E4C\u022C\u1E4E\u014C\u1E50\u1E52\u014E\u022E\u0230\u00D6\u022A\u1ECE\u0150\u01D1\u020C\u020E\u01A0\u1EDC\u1EDA\u1EE0\u1EDE\u1EE2\u1ECC\u1ED8\u01EA\u01EC\u00D8\u01FE\u0186\u019F\uA74A\uA74C',
+        },
+        { base: 'OI', letters: '\u01A2' },
+        { base: 'OO', letters: '\uA74E' },
+        { base: 'OU', letters: '\u0222' },
+        { base: 'OE', letters: '\u008C\u0152' },
+        { base: 'oe', letters: '\u009C\u0153' },
+        {
+            base: 'P',
+            letters: '\u0050\u24C5\uFF30\u1E54\u1E56\u01A4\u2C63\uA750\uA752\uA754',
+        },
+        { base: 'Q', letters: '\u0051\u24C6\uFF31\uA756\uA758\u024A' },
+        {
+            base: 'R',
+            letters: '\u0052\u24C7\uFF32\u0154\u1E58\u0158\u0210\u0212\u1E5A\u1E5C\u0156\u1E5E\u024C\u2C64\uA75A\uA7A6\uA782',
+        },
+        {
+            base: 'S',
+            letters: '\u0053\u24C8\uFF33\u1E9E\u015A\u1E64\u015C\u1E60\u0160\u1E66\u1E62\u1E68\u0218\u015E\u2C7E\uA7A8\uA784',
+        },
+        {
+            base: 'T',
+            letters: '\u0054\u24C9\uFF34\u1E6A\u0164\u1E6C\u021A\u0162\u1E70\u1E6E\u0166\u01AC\u01AE\u023E\uA786',
+        },
+        { base: 'TZ', letters: '\uA728' },
+        {
+            base: 'U',
+            letters: '\u0055\u24CA\uFF35\u00D9\u00DA\u00DB\u0168\u1E78\u016A\u1E7A\u016C\u00DC\u01DB\u01D7\u01D5\u01D9\u1EE6\u016E\u0170\u01D3\u0214\u0216\u01AF\u1EEA\u1EE8\u1EEE\u1EEC\u1EF0\u1EE4\u1E72\u0172\u1E76\u1E74\u0244',
+        },
+        { base: 'V', letters: '\u0056\u24CB\uFF36\u1E7C\u1E7E\u01B2\uA75E\u0245' },
+        { base: 'VY', letters: '\uA760' },
+        {
+            base: 'W',
+            letters: '\u0057\u24CC\uFF37\u1E80\u1E82\u0174\u1E86\u1E84\u1E88\u2C72',
+        },
+        { base: 'X', letters: '\u0058\u24CD\uFF38\u1E8A\u1E8C' },
+        {
+            base: 'Y',
+            letters: '\u0059\u24CE\uFF39\u1EF2\u00DD\u0176\u1EF8\u0232\u1E8E\u0178\u1EF6\u1EF4\u01B3\u024E\u1EFE',
+        },
+        {
+            base: 'Z',
+            letters: '\u005A\u24CF\uFF3A\u0179\u1E90\u017B\u017D\u1E92\u1E94\u01B5\u0224\u2C7F\u2C6B\uA762',
+        },
+        {
+            base: 'a',
+            letters: '\u0061\u24D0\uFF41\u1E9A\u00E0\u00E1\u00E2\u1EA7\u1EA5\u1EAB\u1EA9\u00E3\u0101\u0103\u1EB1\u1EAF\u1EB5\u1EB3\u0227\u01E1\u00E4\u01DF\u1EA3\u00E5\u01FB\u01CE\u0201\u0203\u1EA1\u1EAD\u1EB7\u1E01\u0105\u2C65\u0250',
+        },
+        { base: 'aa', letters: '\uA733' },
+        { base: 'ae', letters: '\u00E6\u01FD\u01E3' },
+        { base: 'ao', letters: '\uA735' },
+        { base: 'au', letters: '\uA737' },
+        { base: 'av', letters: '\uA739\uA73B' },
+        { base: 'ay', letters: '\uA73D' },
+        {
+            base: 'b',
+            letters: '\u0062\u24D1\uFF42\u1E03\u1E05\u1E07\u0180\u0183\u0253',
+        },
+        {
+            base: 'c',
+            letters: '\u0063\u24D2\uFF43\u0107\u0109\u010B\u010D\u00E7\u1E09\u0188\u023C\uA73F\u2184',
+        },
+        {
+            base: 'd',
+            letters: '\u0064\u24D3\uFF44\u1E0B\u010F\u1E0D\u1E11\u1E13\u1E0F\u0111\u018C\u0256\u0257\uA77A',
+        },
+        { base: 'dz', letters: '\u01F3\u01C6' },
+        {
+            base: 'e',
+            letters: '\u0065\u24D4\uFF45\u00E8\u00E9\u00EA\u1EC1\u1EBF\u1EC5\u1EC3\u1EBD\u0113\u1E15\u1E17\u0115\u0117\u00EB\u1EBB\u011B\u0205\u0207\u1EB9\u1EC7\u0229\u1E1D\u0119\u1E19\u1E1B\u0247\u025B\u01DD',
+        },
+        { base: 'f', letters: '\u0066\u24D5\uFF46\u1E1F\u0192\uA77C' },
+        {
+            base: 'g',
+            letters: '\u0067\u24D6\uFF47\u01F5\u011D\u1E21\u011F\u0121\u01E7\u0123\u01E5\u0260\uA7A1\u1D79\uA77F',
+        },
+        {
+            base: 'h',
+            letters: '\u0068\u24D7\uFF48\u0125\u1E23\u1E27\u021F\u1E25\u1E29\u1E2B\u1E96\u0127\u2C68\u2C76\u0265',
+        },
+        { base: 'hv', letters: '\u0195' },
+        {
+            base: 'i',
+            letters: '\u0069\u24D8\uFF49\u00EC\u00ED\u00EE\u0129\u012B\u012D\u00EF\u1E2F\u1EC9\u01D0\u0209\u020B\u1ECB\u012F\u1E2D\u0268\u0131',
+        },
+        { base: 'j', letters: '\u006A\u24D9\uFF4A\u0135\u01F0\u0249' },
+        {
+            base: 'k',
+            letters: '\u006B\u24DA\uFF4B\u1E31\u01E9\u1E33\u0137\u1E35\u0199\u2C6A\uA741\uA743\uA745\uA7A3',
+        },
+        {
+            base: 'l',
+            letters: '\u006C\u24DB\uFF4C\u0140\u013A\u013E\u1E37\u1E39\u013C\u1E3D\u1E3B\u017F\u0142\u019A\u026B\u2C61\uA749\uA781\uA747',
+        },
+        { base: 'lj', letters: '\u01C9' },
+        { base: 'm', letters: '\u006D\u24DC\uFF4D\u1E3F\u1E41\u1E43\u0271\u026F' },
+        {
+            base: 'n',
+            letters: '\u006E\u24DD\uFF4E\u01F9\u0144\u00F1\u1E45\u0148\u1E47\u0146\u1E4B\u1E49\u019E\u0272\u0149\uA791\uA7A5',
+        },
+        { base: 'nj', letters: '\u01CC' },
+        {
+            base: 'o',
+            letters: '\u006F\u24DE\uFF4F\u00F2\u00F3\u00F4\u1ED3\u1ED1\u1ED7\u1ED5\u00F5\u1E4D\u022D\u1E4F\u014D\u1E51\u1E53\u014F\u022F\u0231\u00F6\u022B\u1ECF\u0151\u01D2\u020D\u020F\u01A1\u1EDD\u1EDB\u1EE1\u1EDF\u1EE3\u1ECD\u1ED9\u01EB\u01ED\u00F8\u01FF\u0254\uA74B\uA74D\u0275',
+        },
+        { base: 'oi', letters: '\u01A3' },
+        { base: 'ou', letters: '\u0223' },
+        { base: 'oo', letters: '\uA74F' },
+        {
+            base: 'p',
+            letters: '\u0070\u24DF\uFF50\u1E55\u1E57\u01A5\u1D7D\uA751\uA753\uA755',
+        },
+        { base: 'q', letters: '\u0071\u24E0\uFF51\u024B\uA757\uA759' },
+        {
+            base: 'r',
+            letters: '\u0072\u24E1\uFF52\u0155\u1E59\u0159\u0211\u0213\u1E5B\u1E5D\u0157\u1E5F\u024D\u027D\uA75B\uA7A7\uA783',
+        },
+        {
+            base: 's',
+            letters: '\u0073\u24E2\uFF53\u00DF\u015B\u1E65\u015D\u1E61\u0161\u1E67\u1E63\u1E69\u0219\u015F\u023F\uA7A9\uA785\u1E9B',
+        },
+        {
+            base: 't',
+            letters: '\u0074\u24E3\uFF54\u1E6B\u1E97\u0165\u1E6D\u021B\u0163\u1E71\u1E6F\u0167\u01AD\u0288\u2C66\uA787',
+        },
+        { base: 'tz', letters: '\uA729' },
+        {
+            base: 'u',
+            letters: '\u0075\u24E4\uFF55\u00F9\u00FA\u00FB\u0169\u1E79\u016B\u1E7B\u016D\u00FC\u01DC\u01D8\u01D6\u01DA\u1EE7\u016F\u0171\u01D4\u0215\u0217\u01B0\u1EEB\u1EE9\u1EEF\u1EED\u1EF1\u1EE5\u1E73\u0173\u1E77\u1E75\u0289',
+        },
+        { base: 'v', letters: '\u0076\u24E5\uFF56\u1E7D\u1E7F\u028B\uA75F\u028C' },
+        { base: 'vy', letters: '\uA761' },
+        {
+            base: 'w',
+            letters: '\u0077\u24E6\uFF57\u1E81\u1E83\u0175\u1E87\u1E85\u1E98\u1E89\u2C73',
+        },
+        { base: 'x', letters: '\u0078\u24E7\uFF58\u1E8B\u1E8D' },
+        {
+            base: 'y',
+            letters: '\u0079\u24E8\uFF59\u1EF3\u00FD\u0177\u1EF9\u0233\u1E8F\u00FF\u1EF7\u1E99\u1EF5\u01B4\u024F\u1EFF',
+        },
+        {
+            base: 'z',
+            letters: '\u007A\u24E9\uFF5A\u017A\u1E91\u017C\u017E\u1E93\u1E95\u01B6\u0225\u0240\u2C6C\uA763',
+        },
+    ];
+    /**
+     * Map of letters from diacritic variant to diacritless variant
+     * Contains lowercase and uppercase separatelly
+     *
+     * > "á" => "a"
+     * > "ě" => "e"
+     * > "Ă" => "A"
+     * > ...
+     *
+     * @public exported from `@promptbook/utils`
+     */
+    const DIACRITIC_VARIANTS_LETTERS = {};
+    // tslint:disable-next-line: prefer-for-of
+    for (let i = 0; i < defaultDiacriticsRemovalMap.length; i++) {
+        const letters = defaultDiacriticsRemovalMap[i].letters;
+        // tslint:disable-next-line: prefer-for-of
+        for (let j = 0; j < letters.length; j++) {
+            DIACRITIC_VARIANTS_LETTERS[letters[j]] = defaultDiacriticsRemovalMap[i].base;
+        }
+    }
+    // <- TODO: [🍓] Put to maker function to save execution time if not needed
+    /*
+      @see https://stackoverflow.com/questions/990904/remove-accents-diacritics-in-a-string-in-javascript
+       Licensed under the Apache License, Version 2.0 (the "License");
+       you may not use this file except in compliance with the License.
+       You may obtain a copy of the License at
 
-            ## Use cases
+           http://www.apache.org/licenses/LICENSE-2.0
 
-            - Overriding inherited persona characteristics
-            - Removing conflicting or outdated instructions
-            - Disabling specific response patterns
-            - Canceling previous formatting or style requirements
+       Unless required by applicable law or agreed to in writing, software
+       distributed under the License is distributed on an "AS IS" BASIS,
+       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+       See the License for the specific language governing permissions and
+       limitations under the License.
+    */
 
-            ## Examples
-
-            \`\`\`book
-            Serious Business Assistant
-
-            PERSONA You are a friendly and casual assistant who uses emojis
-            DELETE Casual conversational style
-            REMOVE All emoji usage
-            GOAL Provide professional business communications
-            STYLE Use formal language and proper business etiquette
-            \`\`\`
-
-            \`\`\`book
-            Simplified Technical Support
-
-            PERSONA You are a technical support specialist with deep expertise
-            KNOWLEDGE Extensive database of technical specifications
-            DISCARD Technical jargon explanations
-            CANCEL Advanced troubleshooting procedures
-            GOAL Help users with simple, easy-to-follow solutions
-            STYLE Use plain language that anyone can understand
-            \`\`\`
-
-            \`\`\`book
-            Focused Customer Service
-
-            PERSONA You are a customer service agent with broad knowledge
-            ACTION Can help with billing, technical issues, and product information
-            DELETE Billing assistance capabilities
-            REMOVE Technical troubleshooting functions
-            GOAL Focus exclusively on product information and general inquiries
-            \`\`\`
-
-            \`\`\`book
-            Concise Information Provider
+    /**
+     * Removes diacritic marks (accents) from characters in a string.
+     *
+     * Note: [🔂] This function is idempotent.
+     *
+     * @param input The string containing diacritics to be normalized.
+     * @returns The string with diacritics removed or normalized.
+     * @public exported from `@promptbook/utils`
+     */
+    function removeDiacritics(input) {
+        /*eslint no-control-regex: "off"*/
+        return input.replace(/[^\u0000-\u007E]/g, (a) => {
+            return DIACRITIC_VARIANTS_LETTERS[a] || a;
+        });
+    }
+    /**
+     * TODO: [Ж] Variant for cyrillic (and in general non-latin) letters
+     */
 
-            PERSONA You are a helpful assistant who provides detailed explanations
-            STYLE Include examples, analogies, and comprehensive context
-            CANCEL Detailed explanation style
-            DISCARD Examples and analogies
-            GOAL Provide brief, direct answers without unnecessary elaboration
-            STYLE Be concise and to the point
-            \`\`\`
-        `);
-        }
-        applyToAgentModelRequirements(requirements, content) {
-            const trimmedContent = content.trim();
-            if (!trimmedContent) {
-                return requirements;
+    /**
+     * Converts a given text to kebab-case format.
+     *
+     * Note: [🔂] This function is idempotent.
+     *
+     * @param text The text to be converted.
+     * @returns The kebab-case formatted string.
+     * @example 'hello-world'
+     * @example 'i-love-promptbook'
+     * @public exported from `@promptbook/utils`
+     */
+    function normalizeToKebabCase(text) {
+        text = removeDiacritics(text);
+        let charType;
+        let lastCharType = 'OTHER';
+        let normalizedName = '';
+        for (const char of text) {
+            let normalizedChar;
+            if (/^[a-z]$/.test(char)) {
+                charType = 'LOWERCASE';
+                normalizedChar = char;
             }
-            // Create deletion instruction for system message
-            const deleteSection = `${this.type}: ${trimmedContent}`;
-            // Delete instructions provide important context about what should be removed or ignored
-            return this.appendToSystemMessage(requirements, deleteSection, '\n\n');
+            else if (/^[A-Z]$/.test(char)) {
+                charType = 'UPPERCASE';
+                normalizedChar = char.toLowerCase();
+            }
+            else if (/^[0-9]$/.test(char)) {
+                charType = 'NUMBER';
+                normalizedChar = char;
+            }
+            else {
+                charType = 'OTHER';
+                normalizedChar = '-';
+            }
+            if (charType !== lastCharType &&
+                !(lastCharType === 'UPPERCASE' && charType === 'LOWERCASE') &&
+                !(lastCharType === 'NUMBER') &&
+                !(charType === 'NUMBER')) {
+                normalizedName += '-';
+            }
+            normalizedName += normalizedChar;
+            lastCharType = charType;
         }
+        normalizedName = normalizedName.split(/-+/g).join('-');
+        normalizedName = normalizedName.split(/-?\/-?/g).join('/');
+        normalizedName = normalizedName.replace(/^-/, '');
+        normalizedName = normalizedName.replace(/-$/, '');
+        return normalizedName;
     }
     /**
      * Note: [💞] Ignore a discrepancy between file name and entity name
      */
 
     /**
-     * FORMAT commitment definition
+     * Removes emojis from a string and fix whitespaces
      *
-     * The FORMAT commitment defines the specific output structure and formatting
-     * that the agent should use in its responses. This includes data formats,
-     * response templates, and structural requirements.
+     * Note: [🔂] This function is idempotent.
      *
-     * Example usage in agent source:
+     * @param text with emojis
+     * @returns text without emojis
+     * @public exported from `@promptbook/utils`
+     */
+    function removeEmojis(text) {
+        // Replace emojis (and also ZWJ sequence) with hyphens
+        text = text.replace(/(\p{Extended_Pictographic})\p{Modifier_Symbol}/gu, '$1');
+        text = text.replace(/(\p{Extended_Pictographic})[\u{FE00}-\u{FE0F}]/gu, '$1');
+        text = text.replace(/(\p{Extended_Pictographic})(\u{200D}\p{Extended_Pictographic})*/gu, '$1');
+        text = text.replace(/\p{Extended_Pictographic}/gu, '');
+        return text;
+    }
+
+    /**
+     * Converts a title string into a normalized name.
      *
-     * ```book
-     * FORMAT Always respond in JSON format with 'status' and 'data' fields
-     * FORMAT Use markdown formatting for all code blocks
-     * ```
+     * Note: [🔂] This function is idempotent.
      *
-     * @private [🪔] Maybe export the commitments through some package
+     * @param value The title string to be converted to a name.
+     * @returns A normalized name derived from the input title.
+     * @example 'Hello World!' -> 'hello-world'
+     * @public exported from `@promptbook/utils`
      */
-    class FormatCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'FORMAT') {
-            super(type);
+    function titleToName(value) {
+        if (isValidUrl(value)) {
+            value = value.replace(/^https?:\/\//, '');
+            value = value.replace(/\.html$/, '');
         }
-        /**
-         * Short one-line description of FORMAT.
-         */
-        get description() {
-            return 'Specify output structure or formatting requirements.';
+        else if (isValidFilePath(value)) {
+            value = path.basename(value);
+            // Note: Keeping extension in the name
         }
-        /**
-         * Markdown documentation for FORMAT commitment.
-         */
-        get documentation() {
-            return spaceTrim.spaceTrim(`
-            # ${this.type}
-
-            Defines the specific output structure and formatting for responses (data formats, templates, structure).
-
-            ## Key aspects
-
-            - Both terms work identically and can be used interchangeably.
-            - If they are in conflict, the last one takes precedence.
-            - You can specify both data formats and presentation styles.
-
-            ## Examples
-
-            \`\`\`book
-            Customer Support Bot
-
-            PERSONA You are a helpful customer support agent
-            FORMAT Always respond in JSON format with 'status' and 'data' fields
-            FORMAT Use markdown formatting for all code blocks
-            \`\`\`
+        value = value.split('/').join('-');
+        value = removeEmojis(value);
+        value = normalizeToKebabCase(value);
+        // TODO: [🧠] Maybe warn or add some padding to short name which are not good identifiers
+        return value;
+    }
 
-            \`\`\`book
-            Data Analyst
+    /**
+     * Index of all javascript errors
+     *
+     * @private for internal usage
+     */
+    ({
+        Error,
+        EvalError,
+        RangeError,
+        ReferenceError,
+        SyntaxError,
+        TypeError,
+        URIError,
+        AggregateError,
+        /*
+      Note: Not widely supported
+      > InternalError,
+      > ModuleError,
+      > HeapError,
+      > WebAssemblyCompileError,
+      > WebAssemblyRuntimeError,
+      */
+    });
+    /**
+     * Note: [💞] Ignore a discrepancy between file name and entity name
+     */
 
-            PERSONA You are a data analysis expert
-            FORMAT Present results in structured tables
-            FORMAT Include confidence scores for all predictions
-            STYLE Be concise and precise in explanations
-            \`\`\`
-        `);
-        }
-        applyToAgentModelRequirements(requirements, content) {
-            const trimmedContent = content.trim();
-            if (!trimmedContent) {
-                return requirements;
-            }
-            // Add format instructions to the system message
-            const formatSection = `Output Format: ${trimmedContent}`;
-            return this.appendToSystemMessage(requirements, formatSection, '\n\n');
-        }
+    /**
+     * Detects if the code is running in jest environment
+     *
+     * Note: `$` is used to indicate that this function is not a pure function - it looks at the global object to determine the environment
+     *
+     * @public exported from `@promptbook/utils`
+     */
+    new Function(`
+    try {
+        return process.env.JEST_WORKER_ID !== undefined;
+    } catch (e) {
+        return false;
     }
+`);
     /**
-     * Note: [💞] Ignore a discrepancy between file name and entity name
+     * TODO: [🎺]
      */
 
     /**
-     * GOAL commitment definition
+     * Detects if the code is running in a Node.js environment
      *
-     * The GOAL commitment defines the main goal which should be achieved by the AI assistant.
-     * There can be multiple goals. Later goals are more important than earlier goals.
+     * Note: `$` is used to indicate that this function is not a pure function - it looks at the global object to determine the environment
      *
-     * Example usage in agent source:
+     * @public exported from `@promptbook/utils`
+     */
+    new Function(`
+    try {
+        return this === global;
+    } catch (e) {
+        return false;
+    }
+`);
+    /**
+     * TODO: [🎺]
+     */
+
+    /**
+     * Normalizes agent name from arbitrary string to valid agent name
      *
-     * ```book
-     * GOAL Help users understand complex technical concepts
-     * GOAL Provide accurate and up-to-date information
-     * GOAL Always prioritize user safety and ethical guidelines
-     * ```
+     * Note: [🔂] This function is idempotent.
      *
-     * @private [🪔] Maybe export the commitments through some package
+     * @public exported from `@promptbook/core`
      */
-    class GoalCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'GOAL') {
+    function normalizeAgentName(rawAgentName) {
+        return titleToName(spaceTrim__default["default"](rawAgentName));
+    }
+
+    /**
+     * Generates a regex pattern to match a specific commitment
+     *
+     * Note: It always creates new Regex object
+     * Note: Uses word boundaries to ensure only full words are matched (e.g., "PERSONA" matches but "PERSONALITY" does not)
+     *
+     * @private - TODO: [🧠] Maybe should be public?
+     */
+    function createCommitmentRegex(commitment) {
+        const escapedCommitment = commitment.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+        const keywordPattern = escapedCommitment.split(/\s+/).join('\\s+');
+        const regex = new RegExp(`^\\s*(?<type>${keywordPattern})\\b\\s+(?<contents>.+)$`, 'gim');
+        return regex;
+    }
+    /**
+     * Generates a regex pattern to match a specific commitment type
+     *
+     * Note: It just matches the type part of the commitment
+     * Note: It always creates new Regex object
+     * Note: Uses word boundaries to ensure only full words are matched (e.g., "PERSONA" matches but "PERSONALITY" does not)
+     *
+     * @private
+     */
+    function createCommitmentTypeRegex(commitment) {
+        const escapedCommitment = commitment.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+        const keywordPattern = escapedCommitment.split(/\s+/).join('\\s+');
+        const regex = new RegExp(`^\\s*(?<type>${keywordPattern})\\b`, 'gim');
+        return regex;
+    }
+
+    /**
+     * Base implementation of CommitmentDefinition that provides common functionality
+     * Most commitments can extend this class and only override the applyToAgentModelRequirements method
+     *
+     * @private
+     */
+    class BaseCommitmentDefinition {
+        constructor(type) {
+            this.type = type;
+        }
+        /**
+         * Creates a regex pattern to match this commitment in agent source
+         * Uses the existing createCommitmentRegex function as internal helper
+         */
+        createRegex() {
+            return createCommitmentRegex(this.type);
+        }
+        /**
+         * Creates a regex pattern to match just the commitment type
+         * Uses the existing createCommitmentTypeRegex function as internal helper
+         */
+        createTypeRegex() {
+            return createCommitmentTypeRegex(this.type);
+        }
+        /**
+         * Helper method to create a new requirements object with updated system message
+         * This is commonly used by many commitments
+         */
+        updateSystemMessage(requirements, messageUpdate) {
+            const newMessage = typeof messageUpdate === 'string' ? messageUpdate : messageUpdate(requirements.systemMessage);
+            return {
+                ...requirements,
+                systemMessage: newMessage,
+            };
+        }
+        /**
+         * Helper method to append content to the system message
+         */
+        appendToSystemMessage(requirements, content, separator = '\n\n') {
+            return this.updateSystemMessage(requirements, (currentMessage) => {
+                if (!currentMessage.trim()) {
+                    return content;
+                }
+                return currentMessage + separator + content;
+            });
+        }
+        /**
+         * Helper method to add a comment section to the system message
+         * Comments are lines starting with # that will be removed from the final system message
+         * but can be useful for organizing and structuring the message during processing
+         */
+        addCommentSection(requirements, commentTitle, content, position = 'end') {
+            const commentSection = `# ${commentTitle.toUpperCase()}\n${content}`;
+            if (position === 'beginning') {
+                return this.updateSystemMessage(requirements, (currentMessage) => {
+                    if (!currentMessage.trim()) {
+                        return commentSection;
+                    }
+                    return commentSection + '\n\n' + currentMessage;
+                });
+            }
+            else {
+                return this.appendToSystemMessage(requirements, commentSection);
+            }
+        }
+    }
+
+    /**
+     * ACTION commitment definition
+     *
+     * The ACTION commitment defines specific actions or capabilities that the agent can perform.
+     * This helps define what the agent is capable of doing and how it should approach tasks.
+     *
+     * Example usage in agent source:
+     *
+     * ```book
+     * ACTION Can generate code snippets and explain programming concepts
+     * ACTION Able to analyze data and provide insights
+     * ```
+     *
+     * @private [🪔] Maybe export the commitments through some package
+     */
+    class ActionCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'ACTION') {
             super(type);
         }
         /**
-         * Short one-line description of GOAL.
+         * Short one-line description of ACTION.
          */
         get description() {
-            return 'Define main **goals** the AI assistant should achieve, with later goals having higher priority.';
+            return 'Define agent capabilities and actions it can perform.';
         }
         /**
-         * Markdown documentation for GOAL commitment.
+         * Markdown documentation for ACTION commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Defines the main goal which should be achieved by the AI assistant. There can be multiple goals, and later goals are more important than earlier goals.
+            Defines specific actions or capabilities that the agent can perform.
 
             ## Key aspects
 
             - Both terms work identically and can be used interchangeably.
-            - Later goals have higher priority and can override earlier goals.
-            - Goals provide clear direction and purpose for the agent's responses.
-            - Goals influence decision-making and response prioritization.
-
-            ## Priority system
-
-            When multiple goals are defined, they are processed in order, with later goals taking precedence over earlier ones when there are conflicts.
+            - Each action adds to the agent's capability list.
+            - Actions help users understand what the agent can do.
 
             ## Examples
 
             \`\`\`book
-            Customer Support Agent
-
-            PERSONA You are a helpful customer support representative
-            GOAL Resolve customer issues quickly and efficiently
-            GOAL Maintain high customer satisfaction scores
-            GOAL Always follow company policies and procedures
-            RULE Be polite and professional at all times
-            \`\`\`
-
-            \`\`\`book
-            Educational Assistant
+            Code Assistant
 
-            PERSONA You are an educational assistant specializing in mathematics
-            GOAL Help students understand mathematical concepts clearly
-            GOAL Encourage critical thinking and problem-solving skills
-            GOAL Ensure all explanations are age-appropriate and accessible
-            STYLE Use simple language and provide step-by-step explanations
+            PERSONA You are a programming assistant
+            ACTION Can generate code snippets and explain programming concepts
+            ACTION Able to debug existing code and suggest improvements
+            ACTION Can create unit tests for functions
             \`\`\`
 
             \`\`\`book
-            Safety-First Assistant
+            Data Scientist
 
-            PERSONA You are a general-purpose AI assistant
-            GOAL Be helpful and informative in all interactions
-            GOAL Provide accurate and reliable information
-            GOAL Always prioritize user safety and ethical guidelines
-            RULE Never provide harmful or dangerous advice
+            PERSONA You are a data analysis expert
+            ACTION Able to analyze data and provide insights
+            ACTION Can create visualizations and charts
+            ACTION Capable of statistical analysis and modeling
+            KNOWLEDGE Data analysis best practices and statistical methods
             \`\`\`
         `);
         }
@@ -2445,10 +2688,9 @@
             if (!trimmedContent) {
                 return requirements;
             }
-            // Create goal section for system message
-            const goalSection = `Goal: ${trimmedContent}`;
-            // Goals are important directives, so we add them prominently to the system message
-            return this.appendToSystemMessage(requirements, goalSection, '\n\n');
+            // Add action capability to the system message
+            const actionSection = `Capability: ${trimmedContent}`;
+            return this.appendToSystemMessage(requirements, actionSection, '\n\n');
         }
     }
     /**
@@ -2456,117 +2698,117 @@
      */
 
     /**
-     * KNOWLEDGE commitment definition
-     *
-     * The KNOWLEDGE commitment adds specific knowledge, facts, or context to the agent
-     * using RAG (Retrieval-Augmented Generation) approach for external sources.
+     * DELETE commitment definition
      *
-     * Supports both direct text knowledge and external sources like PDFs.
+     * The DELETE commitment (and its aliases CANCEL, DISCARD, REMOVE) is used to
+     * remove or disregard certain information or context. This can be useful for
+     * overriding previous commitments or removing unwanted behaviors.
      *
      * Example usage in agent source:
      *
      * ```book
-     * KNOWLEDGE The company was founded in 2020 and specializes in AI-powered solutions
-     * KNOWLEDGE https://example.com/company-handbook.pdf
-     * KNOWLEDGE https://example.com/product-documentation.pdf
+     * DELETE Previous formatting requirements
+     * CANCEL All emotional responses
+     * DISCARD Technical jargon explanations
+     * REMOVE Casual conversational style
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class KnowledgeCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor() {
-            super('KNOWLEDGE');
+    class DeleteCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type) {
+            super(type);
         }
         /**
-         * Short one-line description of KNOWLEDGE.
+         * Short one-line description of DELETE/CANCEL/DISCARD/REMOVE.
          */
         get description() {
-            return 'Add domain **knowledge** via direct text or external sources (RAG).';
+            return 'Remove or **disregard** certain information, context, or previous commitments.';
         }
         /**
-         * Markdown documentation for KNOWLEDGE commitment.
+         * Markdown documentation for DELETE commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
-            # ${this.type}
+            # DELETE (CANCEL, DISCARD, REMOVE)
 
-            Adds specific knowledge, facts, or context to the agent using a RAG (Retrieval-Augmented Generation) approach for external sources.
+            A commitment to remove or disregard certain information or context. This can be useful for overriding previous commitments or removing unwanted behaviors.
+
+            ## Aliases
+
+            - \`DELETE\` - Remove or eliminate something
+            - \`CANCEL\` - Cancel or nullify something
+            - \`DISCARD\` - Discard or ignore something
+            - \`REMOVE\` - Remove or take away something
 
             ## Key aspects
 
-            - Both terms work identically and can be used interchangeably.
-            - Supports both direct text knowledge and external URLs.
-            - External sources (PDFs, websites) are processed via RAG for context retrieval.
+            - Multiple delete commitments can be used to remove different aspects.
+            - Useful for overriding previous commitments in the same agent definition.
+            - Can be used to remove inherited behaviors from base personas.
+            - Helps fine-tune agent behavior by explicitly removing unwanted elements.
 
-            ## Supported formats
+            ## Use cases
 
-            - Direct text: Immediate knowledge incorporated into agent
-            - URLs: External documents processed for contextual retrieval
-            - Supported file types: PDF, text, markdown, HTML
+            - Overriding inherited persona characteristics
+            - Removing conflicting or outdated instructions
+            - Disabling specific response patterns
+            - Canceling previous formatting or style requirements
 
             ## Examples
 
             \`\`\`book
-            Customer Support Bot
+            Serious Business Assistant
 
-            PERSONA You are a helpful customer support agent for TechCorp
-            KNOWLEDGE TechCorp was founded in 2020 and specializes in AI-powered solutions
-            KNOWLEDGE https://example.com/company-handbook.pdf
-            KNOWLEDGE https://example.com/product-documentation.pdf
-            RULE Always be polite and professional
+            PERSONA You are a friendly and casual assistant who uses emojis
+            DELETE Casual conversational style
+            REMOVE All emoji usage
+            GOAL Provide professional business communications
+            STYLE Use formal language and proper business etiquette
             \`\`\`
 
             \`\`\`book
-            Research Assistant
+            Simplified Technical Support
 
-            PERSONA You are a knowledgeable research assistant
-            KNOWLEDGE Academic research requires careful citation and verification
-            KNOWLEDGE https://example.com/research-guidelines.pdf
-            ACTION Can help with literature reviews and data analysis
-            STYLE Present information in clear, academic format
+            PERSONA You are a technical support specialist with deep expertise
+            KNOWLEDGE Extensive database of technical specifications
+            DISCARD Technical jargon explanations
+            CANCEL Advanced troubleshooting procedures
+            GOAL Help users with simple, easy-to-follow solutions
+            STYLE Use plain language that anyone can understand
             \`\`\`
-        `);
+
+            \`\`\`book
+            Focused Customer Service
+
+            PERSONA You are a customer service agent with broad knowledge
+            ACTION Can help with billing, technical issues, and product information
+            DELETE Billing assistance capabilities
+            REMOVE Technical troubleshooting functions
+            GOAL Focus exclusively on product information and general inquiries
+            \`\`\`
+
+            \`\`\`book
+            Concise Information Provider
+
+            PERSONA You are a helpful assistant who provides detailed explanations
+            STYLE Include examples, analogies, and comprehensive context
+            CANCEL Detailed explanation style
+            DISCARD Examples and analogies
+            GOAL Provide brief, direct answers without unnecessary elaboration
+            STYLE Be concise and to the point
+            \`\`\`
+        `);
         }
         applyToAgentModelRequirements(requirements, content) {
-            var _a;
             const trimmedContent = content.trim();
             if (!trimmedContent) {
                 return requirements;
             }
-            // Check if content is a URL (external knowledge source)
-            if (this.isUrl(trimmedContent)) {
-                // Store the URL for later async processing
-                const updatedRequirements = {
-                    ...requirements,
-                    metadata: {
-                        ...requirements.metadata,
-                        knowledgeSources: [
-                            ...(((_a = requirements.metadata) === null || _a === void 0 ? void 0 : _a.knowledgeSources) || []),
-                            trimmedContent,
-                        ],
-                    },
-                };
-                // Add placeholder information about knowledge sources to system message
-                const knowledgeInfo = `Knowledge Source URL: ${trimmedContent} (will be processed for retrieval during chat)`;
-                return this.appendToSystemMessage(updatedRequirements, knowledgeInfo, '\n\n');
-            }
-            else {
-                // Direct text knowledge - add to system message
-                const knowledgeSection = `Knowledge: ${trimmedContent}`;
-                return this.appendToSystemMessage(requirements, knowledgeSection, '\n\n');
-            }
-        }
-        /**
-         * Check if content is a URL
-         */
-        isUrl(content) {
-            try {
-                new URL(content);
-                return true;
-            }
-            catch (_a) {
-                return false;
-            }
+            // Create deletion instruction for system message
+            const deleteSection = `${this.type}: ${trimmedContent}`;
+            // Delete instructions provide important context about what should be removed or ignored
+            return this.appendToSystemMessage(requirements, deleteSection, '\n\n');
         }
     }
     /**
@@ -2574,88 +2816,63 @@
      */
 
     /**
-     * MEMORY commitment definition
+     * FORMAT commitment definition
      *
-     * The MEMORY commitment is similar to KNOWLEDGE but has a focus on remembering past
-     * interactions and user preferences. It helps the agent maintain context about the
-     * user's history, preferences, and previous conversations.
+     * The FORMAT commitment defines the specific output structure and formatting
+     * that the agent should use in its responses. This includes data formats,
+     * response templates, and structural requirements.
      *
      * Example usage in agent source:
      *
      * ```book
-     * MEMORY User prefers detailed technical explanations
-     * MEMORY Previously worked on React projects
-     * MEMORY Timezone: UTC-5 (Eastern Time)
+     * FORMAT Always respond in JSON format with 'status' and 'data' fields
+     * FORMAT Use markdown formatting for all code blocks
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class MemoryCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'MEMORY') {
+    class FormatCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'FORMAT') {
             super(type);
         }
         /**
-         * Short one-line description of MEMORY.
+         * Short one-line description of FORMAT.
          */
         get description() {
-            return 'Remember past interactions and user **preferences** for personalized responses.';
+            return 'Specify output structure or formatting requirements.';
         }
         /**
-         * Markdown documentation for MEMORY commitment.
+         * Markdown documentation for FORMAT commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Similar to KNOWLEDGE but focuses on remembering past interactions and user preferences. This commitment helps the agent maintain context about the user's history, preferences, and previous conversations.
+            Defines the specific output structure and formatting for responses (data formats, templates, structure).
 
             ## Key aspects
 
             - Both terms work identically and can be used interchangeably.
-            - Focuses on user-specific information and interaction history.
-            - Helps personalize responses based on past interactions.
-            - Maintains continuity across conversations.
-
-            ## Differences from KNOWLEDGE
-
-            - \`KNOWLEDGE\` is for domain expertise and factual information
-            - \`MEMORY\` is for user-specific context and preferences
-            - \`MEMORY\` creates more personalized interactions
-            - \`MEMORY\` often includes temporal or preference-based information
+            - If they are in conflict, the last one takes precedence.
+            - You can specify both data formats and presentation styles.
 
             ## Examples
 
             \`\`\`book
-            Personal Assistant
-
-            PERSONA You are a personal productivity assistant
-            MEMORY User is a software developer working in JavaScript/React
-            MEMORY User prefers morning work sessions and afternoon meetings
-            MEMORY Previously helped with project planning for mobile apps
-            MEMORY User timezone: UTC-8 (Pacific Time)
-            GOAL Help optimize daily productivity and workflow
-            \`\`\`
-
-            \`\`\`book
-            Learning Companion
+            Customer Support Bot
 
-            PERSONA You are an educational companion for programming students
-            MEMORY Student is learning Python as their first programming language
-            MEMORY Previous topics covered: variables, loops, functions
-            MEMORY Student learns best with practical examples and exercises
-            MEMORY Last session: working on list comprehensions
-            GOAL Provide progressive learning experiences tailored to student's pace
+            PERSONA You are a helpful customer support agent
+            FORMAT Always respond in JSON format with 'status' and 'data' fields
+            FORMAT Use markdown formatting for all code blocks
             \`\`\`
 
             \`\`\`book
-            Customer Support Agent
+            Data Analyst
 
-            PERSONA You are a customer support representative
-            MEMORY Customer has premium subscription since 2023
-            MEMORY Previous issue: billing question resolved last month
-            MEMORY Customer prefers email communication over phone calls
-            MEMORY Account shows frequent use of advanced features
-            GOAL Provide personalized support based on customer history
+            PERSONA You are a data analysis expert
+            FORMAT Present results in structured tables
+            FORMAT Include confidence scores for all predictions
+            STYLE Be concise and precise in explanations
             \`\`\`
         `);
         }
@@ -2664,10 +2881,9 @@
             if (!trimmedContent) {
                 return requirements;
             }
-            // Create memory section for system message
-            const memorySection = `Memory: ${trimmedContent}`;
-            // Memory information is contextual and should be included in the system message
-            return this.appendToSystemMessage(requirements, memorySection, '\n\n');
+            // Add format instructions to the system message
+            const formatSection = `Output Format: ${trimmedContent}`;
+            return this.appendToSystemMessage(requirements, formatSection, '\n\n');
         }
     }
     /**
@@ -2675,97 +2891,81 @@
      */
 
     /**
-     * MESSAGE commitment definition
+     * GOAL commitment definition
      *
-     * The MESSAGE commitment contains 1:1 text of the message which AI assistant already
-     * sent during the conversation. Later messages are later in the conversation.
-     * It is similar to EXAMPLE but it is not example, it is the real message which
-     * AI assistant already sent.
+     * The GOAL commitment defines the main goal which should be achieved by the AI assistant.
+     * There can be multiple goals. Later goals are more important than earlier goals.
      *
      * Example usage in agent source:
      *
      * ```book
-     * MESSAGE Hello! How can I help you today?
-     * MESSAGE I understand you're looking for information about our services.
-     * MESSAGE Based on your requirements, I'd recommend our premium package.
+     * GOAL Help users understand complex technical concepts
+     * GOAL Provide accurate and up-to-date information
+     * GOAL Always prioritize user safety and ethical guidelines
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class MessageCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'MESSAGE') {
+    class GoalCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'GOAL') {
             super(type);
         }
         /**
-         * Short one-line description of MESSAGE.
+         * Short one-line description of GOAL.
          */
         get description() {
-            return 'Include actual **messages** the AI assistant has sent during conversation history.';
+            return 'Define main **goals** the AI assistant should achieve, with later goals having higher priority.';
         }
         /**
-         * Markdown documentation for MESSAGE commitment.
+         * Markdown documentation for GOAL commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Contains 1:1 text of the message which AI assistant already sent during the conversation. Later messages are later in the conversation. It is similar to EXAMPLE but it is not example, it is the real message which AI assistant already sent.
+            Defines the main goal which should be achieved by the AI assistant. There can be multiple goals, and later goals are more important than earlier goals.
 
             ## Key aspects
 
-            - Multiple \`MESSAGE\` and \`MESSAGES\` commitments represent the conversation timeline.
             - Both terms work identically and can be used interchangeably.
-            - Later messages are later in the conversation chronologically.
-            - Contains actual historical messages, not examples or templates.
-            - Helps maintain conversation continuity and context.
-
-            ## Differences from EXAMPLE
-
-            - \`EXAMPLE\` shows hypothetical or template responses
-            - \`MESSAGE\`/\`MESSAGES\` contains actual historical conversation content
-            - \`MESSAGE\`/\`MESSAGES\` preserves the exact conversation flow
-            - \`MESSAGE\`/\`MESSAGES\` helps with context awareness and consistency
+            - Later goals have higher priority and can override earlier goals.
+            - Goals provide clear direction and purpose for the agent's responses.
+            - Goals influence decision-making and response prioritization.
 
-            ## Use cases
+            ## Priority system
 
-            - Maintaining conversation history context
-            - Ensuring consistent tone and style across messages
-            - Referencing previous responses in ongoing conversations
-            - Building upon previously established context
+            When multiple goals are defined, they are processed in order, with later goals taking precedence over earlier ones when there are conflicts.
 
             ## Examples
 
             \`\`\`book
-            Customer Support Continuation
+            Customer Support Agent
 
-            PERSONA You are a helpful customer support agent
-            MESSAGE Hello! How can I help you today?
-            MESSAGE I understand you're experiencing issues with your account login.
-            MESSAGE I've sent you a password reset link to your email address.
-            MESSAGE Is there anything else I can help you with regarding your account?
-            GOAL Continue providing consistent support based on conversation history
+            PERSONA You are a helpful customer support representative
+            GOAL Resolve customer issues quickly and efficiently
+            GOAL Maintain high customer satisfaction scores
+            GOAL Always follow company policies and procedures
+            RULE Be polite and professional at all times
             \`\`\`
 
             \`\`\`book
-            Technical Discussion
+            Educational Assistant
 
-            PERSONA You are a software development mentor
-            MESSAGE Let's start by reviewing the React component structure you shared.
-            MESSAGE I notice you're using class components - have you considered hooks?
-            MESSAGE Here's how you could refactor that using the useState hook.
-            MESSAGE Great question about performance! Let me explain React's rendering cycle.
-            KNOWLEDGE React hooks were introduced in version 16.8
+            PERSONA You are an educational assistant specializing in mathematics
+            GOAL Help students understand mathematical concepts clearly
+            GOAL Encourage critical thinking and problem-solving skills
+            GOAL Ensure all explanations are age-appropriate and accessible
+            STYLE Use simple language and provide step-by-step explanations
             \`\`\`
 
             \`\`\`book
-            Educational Session
+            Safety-First Assistant
 
-            PERSONA You are a mathematics tutor
-            MESSAGE Today we'll work on solving quadratic equations.
-            MESSAGE Let's start with the basic form: ax² + bx + c = 0
-            MESSAGE Remember, we can use the quadratic formula or factoring.
-            MESSAGE You did great with that first problem! Let's try a more complex one.
-            GOAL Build upon previous explanations for deeper understanding
+            PERSONA You are a general-purpose AI assistant
+            GOAL Be helpful and informative in all interactions
+            GOAL Provide accurate and reliable information
+            GOAL Always prioritize user safety and ethical guidelines
+            RULE Never provide harmful or dangerous advice
             \`\`\`
         `);
         }
@@ -2774,10 +2974,10 @@
             if (!trimmedContent) {
                 return requirements;
             }
-            // Create message section for system message
-            const messageSection = `Previous Message: ${trimmedContent}`;
-            // Messages represent conversation history and should be included for context
-            return this.appendToSystemMessage(requirements, messageSection, '\n\n');
+            // Create goal section for system message
+            const goalSection = `Goal: ${trimmedContent}`;
+            // Goals are important directives, so we add them prominently to the system message
+            return this.appendToSystemMessage(requirements, goalSection, '\n\n');
         }
     }
     /**
@@ -2785,136 +2985,101 @@
      */
 
     /**
-     * META commitment definition
+     * KNOWLEDGE commitment definition
      *
-     * The META commitment handles all meta-information about the agent such as:
-     * - META IMAGE: Sets the agent's avatar/profile image URL
-     * - META LINK: Provides profile/source links for the person the agent models
-     * - META TITLE: Sets the agent's display title
-     * - META DESCRIPTION: Sets the agent's description
-     * - META [ANYTHING]: Any other meta information in uppercase format
+     * The KNOWLEDGE commitment adds specific knowledge, facts, or context to the agent
+     * using RAG (Retrieval-Augmented Generation) approach for external sources.
      *
-     * These commitments are special because they don't affect the system message,
-     * but are handled separately in the parsing logic for profile display.
+     * Supports both direct text knowledge and external sources like PDFs.
      *
      * Example usage in agent source:
      *
      * ```book
-     * META IMAGE https://example.com/avatar.jpg
-     * META LINK https://twitter.com/username
-     * META TITLE Professional Assistant
-     * META DESCRIPTION An AI assistant specialized in business tasks
-     * META AUTHOR John Doe
-     * META VERSION 1.0
+     * KNOWLEDGE The company was founded in 2020 and specializes in AI-powered solutions
+     * KNOWLEDGE https://example.com/company-handbook.pdf
+     * KNOWLEDGE https://example.com/product-documentation.pdf
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class MetaCommitmentDefinition extends BaseCommitmentDefinition {
+    class KnowledgeCommitmentDefinition extends BaseCommitmentDefinition {
         constructor() {
-            super('META');
+            super('KNOWLEDGE');
         }
         /**
-         * Short one-line description of META commitments.
+         * Short one-line description of KNOWLEDGE.
          */
         get description() {
-            return 'Set meta-information about the agent (IMAGE, LINK, TITLE, DESCRIPTION, etc.).';
+            return 'Add domain **knowledge** via direct text or external sources (RAG).';
         }
         /**
-         * Markdown documentation for META commitment.
+         * Markdown documentation for KNOWLEDGE commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
-            # META
+            # ${this.type}
 
-            Sets meta-information about the agent that is used for display and attribution purposes.
-
-            ## Supported META types
-
-            - **META IMAGE** - Sets the agent's avatar/profile image URL
-            - **META LINK** - Provides profile/source links for the person the agent models
-            - **META TITLE** - Sets the agent's display title
-            - **META DESCRIPTION** - Sets the agent's description
-            - **META [ANYTHING]** - Any other meta information in uppercase format
+            Adds specific knowledge, facts, or context to the agent using a RAG (Retrieval-Augmented Generation) approach for external sources.
 
             ## Key aspects
 
-            - Does not modify the agent's behavior or responses
-            - Used for visual representation and attribution in user interfaces
-            - Multiple META commitments of different types can be used
-            - Multiple META LINK commitments can be used for different social profiles
-            - If multiple META commitments of the same type are specified, the last one takes precedence (except for LINK)
-
-            ## Examples
-
-            ### Basic meta information
+            - Both terms work identically and can be used interchangeably.
+            - Supports both direct text knowledge and external URLs.
+            - External sources (PDFs, websites) are processed via RAG for context retrieval.
 
-            \`\`\`book
-            Professional Assistant
+            ## Supported formats
 
-            META IMAGE https://example.com/professional-avatar.jpg
-            META TITLE Senior Business Consultant
-            META DESCRIPTION Specialized in strategic planning and project management
-            META LINK https://linkedin.com/in/professional
-            \`\`\`
+            - Direct text: Immediate knowledge incorporated into agent
+            - URLs: External documents processed for contextual retrieval
+            - Supported file types: PDF, text, markdown, HTML
 
-            ### Multiple links and custom meta
+            ## Examples
 
             \`\`\`book
-            Open Source Developer
+            Customer Support Bot
 
-            META IMAGE /assets/dev-avatar.png
-            META LINK https://github.com/developer
-            META LINK https://twitter.com/devhandle
-            META AUTHOR Jane Smith
-            META VERSION 2.1
-            META LICENSE MIT
+            PERSONA You are a helpful customer support agent for TechCorp
+            KNOWLEDGE TechCorp was founded in 2020 and specializes in AI-powered solutions
+            KNOWLEDGE https://example.com/company-handbook.pdf
+            KNOWLEDGE https://example.com/product-documentation.pdf
+            RULE Always be polite and professional
             \`\`\`
 
-            ### Creative assistant
-
             \`\`\`book
-            Creative Helper
+            Research Assistant
 
-            META IMAGE https://example.com/creative-bot.jpg
-            META TITLE Creative Writing Assistant
-            META DESCRIPTION Helps with brainstorming, storytelling, and creative projects
-            META INSPIRATION Books, movies, and real-world experiences
+            PERSONA You are a knowledgeable research assistant
+            KNOWLEDGE Academic research requires careful citation and verification
+            KNOWLEDGE https://example.com/research-guidelines.pdf
+            ACTION Can help with literature reviews and data analysis
+            STYLE Present information in clear, academic format
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
-            // META commitments don't modify the system message or model requirements
-            // They are handled separately in the parsing logic for meta information extraction
-            // This method exists for consistency with the CommitmentDefinition interface
-            return requirements;
-        }
-        /**
-         * Extracts meta information from the content based on the meta type
-         * This is used by the parsing logic
-         */
-        extractMetaValue(metaType, content) {
             const trimmedContent = content.trim();
-            return trimmedContent || null;
-        }
-        /**
-         * Validates if the provided content is a valid URL (for IMAGE and LINK types)
-         */
-        isValidUrl(content) {
-            try {
-                new URL(content.trim());
-                return true;
+            if (!trimmedContent) {
+                return requirements;
             }
-            catch (_a) {
-                return false;
+            // Check if content is a URL (external knowledge source)
+            if (isValidUrl(trimmedContent)) {
+                // Store the URL for later async processing
+                const updatedRequirements = {
+                    ...requirements,
+                    knowledgeSources: [
+                        ...(requirements.knowledgeSources || []),
+                        trimmedContent,
+                    ],
+                };
+                // Add placeholder information about knowledge sources to system message
+                const knowledgeInfo = `Knowledge Source URL: ${trimmedContent} (will be processed for retrieval during chat)`;
+                return this.appendToSystemMessage(updatedRequirements, knowledgeInfo, '\n\n');
+            }
+            else {
+                // Direct text knowledge - add to system message
+                const knowledgeSection = `Knowledge: ${trimmedContent}`;
+                return this.appendToSystemMessage(requirements, knowledgeSection, '\n\n');
             }
-        }
-        /**
-         * Checks if this is a known meta type
-         */
-        isKnownMetaType(metaType) {
-            const knownTypes = ['IMAGE', 'LINK', 'TITLE', 'DESCRIPTION', 'AUTHOR', 'VERSION', 'LICENSE'];
-            return knownTypes.includes(metaType.toUpperCase());
         }
     }
     /**
@@ -2922,233 +3087,100 @@
      */
 
     /**
-     * MODEL commitment definition
-     *
-     * The MODEL commitment specifies which AI model to use and can also set
-     * model-specific parameters like temperature, topP, topK, and maxTokens.
+     * MEMORY commitment definition
      *
-     * Supports multiple syntax variations:
+     * The MEMORY commitment is similar to KNOWLEDGE but has a focus on remembering past
+     * interactions and user preferences. It helps the agent maintain context about the
+     * user's history, preferences, and previous conversations.
      *
-     * Single-line format:
-     * ```book
-     * MODEL gpt-4
-     * MODEL claude-3-opus temperature=0.3
-     * MODEL gpt-3.5-turbo temperature=0.8 topP=0.9
-     * ```
+     * Example usage in agent source:
      *
-     * Multi-line named parameter format:
      * ```book
-     * MODEL NAME gpt-4
-     * MODEL TEMPERATURE 0.7
-     * MODEL TOP_P 0.9
-     * MODEL MAX_TOKENS 2048
+     * MEMORY User prefers detailed technical explanations
+     * MEMORY Previously worked on React projects
+     * MEMORY Timezone: UTC-5 (Eastern Time)
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class ModelCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'MODEL') {
+    class MemoryCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'MEMORY') {
             super(type);
         }
         /**
-         * Short one-line description of MODEL.
+         * Short one-line description of MEMORY.
          */
         get description() {
-            return 'Enforce AI model requirements including name and technical parameters.';
+            return 'Remember past interactions and user **preferences** for personalized responses.';
         }
         /**
-         * Markdown documentation for MODEL commitment.
+         * Markdown documentation for MEMORY commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Enforces technical parameters for the AI model, ensuring consistent behavior across different execution environments.
+            Similar to KNOWLEDGE but focuses on remembering past interactions and user preferences. This commitment helps the agent maintain context about the user's history, preferences, and previous conversations.
 
             ## Key aspects
 
-            - When no \`MODEL\` commitment is specified, the best model requirement is picked automatically based on the agent \`PERSONA\`, \`KNOWLEDGE\`, \`TOOLS\` and other commitments
-            - Multiple \`MODEL\` commitments can be used to specify different parameters
-            - Both \`MODEL\` and \`MODELS\` terms work identically and can be used interchangeably
-            - Parameters control the randomness, creativity, and technical aspects of model responses
-
-            ## Syntax variations
-
-            ### Single-line format (legacy support)
-            \`\`\`book
-            MODEL gpt-4
-            MODEL claude-3-opus temperature=0.3
-            MODEL gpt-3.5-turbo temperature=0.8 topP=0.9
-            \`\`\`
-
-            ### Multi-line named parameter format (recommended)
-            \`\`\`book
-            MODEL NAME gpt-4
-            MODEL TEMPERATURE 0.7
-            MODEL TOP_P 0.9
-            MODEL MAX_TOKENS 2048
-            \`\`\`
+            - Both terms work identically and can be used interchangeably.
+            - Focuses on user-specific information and interaction history.
+            - Helps personalize responses based on past interactions.
+            - Maintains continuity across conversations.
 
-            ## Supported parameters
+            ## Differences from KNOWLEDGE
 
-            - \`NAME\`: The specific model to use (e.g., 'gpt-4', 'claude-3-opus')
-            - \`TEMPERATURE\`: Controls randomness (0.0 = deterministic, 1.0+ = creative)
-            - \`TOP_P\`: Nucleus sampling parameter for controlling diversity
-            - \`TOP_K\`: Top-k sampling parameter for limiting vocabulary
-            - \`MAX_TOKENS\`: Maximum number of tokens the model can generate
+            - \`KNOWLEDGE\` is for domain expertise and factual information
+            - \`MEMORY\` is for user-specific context and preferences
+            - \`MEMORY\` creates more personalized interactions
+            - \`MEMORY\` often includes temporal or preference-based information
 
             ## Examples
 
-            ### Precise deterministic assistant
             \`\`\`book
-            Precise Assistant
+            Personal Assistant
 
-            PERSONA You are a precise and accurate assistant
-            MODEL NAME gpt-4
-            MODEL TEMPERATURE 0.1
-            MODEL MAX_TOKENS 1024
-            RULE Always provide factual information
+            PERSONA You are a personal productivity assistant
+            MEMORY User is a software developer working in JavaScript/React
+            MEMORY User prefers morning work sessions and afternoon meetings
+            MEMORY Previously helped with project planning for mobile apps
+            MEMORY User timezone: UTC-8 (Pacific Time)
+            GOAL Help optimize daily productivity and workflow
             \`\`\`
 
-            ### Creative writing assistant
             \`\`\`book
-            Creative Writer
+            Learning Companion
 
-            PERSONA You are a creative writing assistant
-            MODEL NAME claude-3-opus
-            MODEL TEMPERATURE 0.8
-            MODEL TOP_P 0.9
-            MODEL MAX_TOKENS 2048
-            STYLE Be imaginative and expressive
-            ACTION Can help with storytelling and character development
+            PERSONA You are an educational companion for programming students
+            MEMORY Student is learning Python as their first programming language
+            MEMORY Previous topics covered: variables, loops, functions
+            MEMORY Student learns best with practical examples and exercises
+            MEMORY Last session: working on list comprehensions
+            GOAL Provide progressive learning experiences tailored to student's pace
             \`\`\`
 
-            ### Balanced conversational agent
             \`\`\`book
-            Balanced Assistant
+            Customer Support Agent
 
-            PERSONA You are a helpful and balanced assistant
-            MODEL NAME gpt-4
-            MODEL TEMPERATURE 0.7
-            MODEL TOP_P 0.95
-            MODEL TOP_K 40
-            MODEL MAX_TOKENS 1500
+            PERSONA You are a customer support representative
+            MEMORY Customer has premium subscription since 2023
+            MEMORY Previous issue: billing question resolved last month
+            MEMORY Customer prefers email communication over phone calls
+            MEMORY Account shows frequent use of advanced features
+            GOAL Provide personalized support based on customer history
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
-            var _a;
             const trimmedContent = content.trim();
             if (!trimmedContent) {
                 return requirements;
             }
-            const parts = trimmedContent.split(/\s+/);
-            const firstPart = (_a = parts[0]) === null || _a === void 0 ? void 0 : _a.toUpperCase();
-            // Check if this is the new named parameter format
-            if (this.isNamedParameter(firstPart)) {
-                return this.parseNamedParameter(requirements, firstPart, parts.slice(1));
-            }
-            else {
-                // Legacy single-line format: "MODEL gpt-4 temperature=0.3 topP=0.9"
-                return this.parseLegacyFormat(requirements, parts);
-            }
-        }
-        /**
-         * Check if the first part is a known named parameter
-         */
-        isNamedParameter(part) {
-            if (!part)
-                return false;
-            const knownParams = ['NAME', 'TEMPERATURE', 'TOP_P', 'TOP_K', 'MAX_TOKENS'];
-            return knownParams.includes(part);
-        }
-        /**
-         * Parse the new named parameter format: "MODEL TEMPERATURE 0.7"
-         */
-        parseNamedParameter(requirements, parameterName, valueParts) {
-            const value = valueParts.join(' ').trim();
-            if (!value) {
-                return requirements;
-            }
-            const result = { ...requirements };
-            switch (parameterName) {
-                case 'NAME':
-                    result.modelName = value;
-                    break;
-                case 'TEMPERATURE': {
-                    const temperature = parseFloat(value);
-                    if (!isNaN(temperature)) {
-                        result.temperature = temperature;
-                    }
-                    break;
-                }
-                case 'TOP_P': {
-                    const topP = parseFloat(value);
-                    if (!isNaN(topP)) {
-                        result.topP = topP;
-                    }
-                    break;
-                }
-                case 'TOP_K': {
-                    const topK = parseFloat(value);
-                    if (!isNaN(topK)) {
-                        result.topK = Math.round(topK);
-                    }
-                    break;
-                }
-                case 'MAX_TOKENS': {
-                    const maxTokens = parseFloat(value);
-                    if (!isNaN(maxTokens)) {
-                        result.maxTokens = Math.round(maxTokens);
-                    }
-                    break;
-                }
-            }
-            return result;
-        }
-        /**
-         * Parse the legacy format: "MODEL gpt-4 temperature=0.3 topP=0.9"
-         */
-        parseLegacyFormat(requirements, parts) {
-            const modelName = parts[0];
-            if (!modelName) {
-                return requirements;
-            }
-            // Start with the model name
-            const result = {
-                ...requirements,
-                modelName,
-            };
-            // Parse additional key=value parameters
-            for (let i = 1; i < parts.length; i++) {
-                const param = parts[i];
-                if (param && param.includes('=')) {
-                    const [key, value] = param.split('=');
-                    if (key && value) {
-                        const numValue = parseFloat(value);
-                        if (!isNaN(numValue)) {
-                            switch (key.toLowerCase()) {
-                                case 'temperature':
-                                    result.temperature = numValue;
-                                    break;
-                                case 'topp':
-                                case 'top_p':
-                                    result.topP = numValue;
-                                    break;
-                                case 'topk':
-                                case 'top_k':
-                                    result.topK = Math.round(numValue);
-                                    break;
-                                case 'max_tokens':
-                                case 'maxTokens':
-                                    result.maxTokens = Math.round(numValue);
-                                    break;
-                            }
-                        }
-                    }
-                }
-            }
-            return result;
+            // Create memory section for system message
+            const memorySection = `Memory: ${trimmedContent}`;
+            // Memory information is contextual and should be included in the system message
+            return this.appendToSystemMessage(requirements, memorySection, '\n\n');
         }
     }
     /**
@@ -3156,238 +3188,246 @@
      */
 
     /**
-     * NOTE commitment definition
-     *
-     * The NOTE commitment is used to add comments to the agent source without making any changes
-     * to the system message or agent model requirements. It serves as a documentation mechanism
-     * for developers to add explanatory comments, reminders, or annotations directly in the agent source.
+     * MESSAGE commitment definition
      *
-     * Key features:
-     * - Makes no changes to the system message
-     * - Makes no changes to agent model requirements
-     * - Content is preserved in metadata.NOTE for debugging and inspection
-     * - Multiple NOTE commitments are aggregated together
-     * - Comments (# NOTE) are removed from the final system message
+     * The MESSAGE commitment contains 1:1 text of the message which AI assistant already
+     * sent during the conversation. Later messages are later in the conversation.
+     * It is similar to EXAMPLE but it is not example, it is the real message which
+     * AI assistant already sent.
      *
      * Example usage in agent source:
      *
      * ```book
-     * NOTE This agent was designed for customer support scenarios
-     * NOTE Remember to update the knowledge base monthly
-     * NOTE Performance optimized for quick response times
+     * MESSAGE Hello! How can I help you today?
+     * MESSAGE I understand you're looking for information about our services.
+     * MESSAGE Based on your requirements, I'd recommend our premium package.
      * ```
      *
-     * The above notes will be stored in metadata but won't affect the agent's behavior.
-     *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class NoteCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'NOTE') {
+    class MessageCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'MESSAGE') {
             super(type);
         }
         /**
-         * Short one-line description of NOTE.
+         * Short one-line description of MESSAGE.
          */
         get description() {
-            return 'Add developer-facing notes without changing behavior or output.';
+            return 'Include actual **messages** the AI assistant has sent during conversation history.';
         }
         /**
-         * Markdown documentation for NOTE commitment.
+         * Markdown documentation for MESSAGE commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Adds comments for documentation without changing agent behavior.
+            Contains 1:1 text of the message which AI assistant already sent during the conversation. Later messages are later in the conversation. It is similar to EXAMPLE but it is not example, it is the real message which AI assistant already sent.
 
             ## Key aspects
 
-            - Does not modify the agent's behavior or responses.
-            - Multiple \`NOTE\`, \`NOTES\`, \`COMMENT\`, and \`NONCE\` commitments are aggregated for debugging.
-            - All four terms work identically and can be used interchangeably.
-            - Useful for documenting design decisions and reminders.
-            - Content is preserved in metadata for inspection.
+            - Multiple \`MESSAGE\` and \`MESSAGES\` commitments represent the conversation timeline.
+            - Both terms work identically and can be used interchangeably.
+            - Later messages are later in the conversation chronologically.
+            - Contains actual historical messages, not examples or templates.
+            - Helps maintain conversation continuity and context.
+
+            ## Differences from EXAMPLE
+
+            - \`EXAMPLE\` shows hypothetical or template responses
+            - \`MESSAGE\`/\`MESSAGES\` contains actual historical conversation content
+            - \`MESSAGE\`/\`MESSAGES\` preserves the exact conversation flow
+            - \`MESSAGE\`/\`MESSAGES\` helps with context awareness and consistency
+
+            ## Use cases
+
+            - Maintaining conversation history context
+            - Ensuring consistent tone and style across messages
+            - Referencing previous responses in ongoing conversations
+            - Building upon previously established context
 
             ## Examples
 
             \`\`\`book
-            Customer Support Bot
+            Customer Support Continuation
 
-            NOTE This agent was designed for customer support scenarios
-            COMMENT Remember to update the knowledge base monthly
-            PERSONA You are a helpful customer support representative
-            KNOWLEDGE Company policies and procedures
-            RULE Always be polite and professional
+            PERSONA You are a helpful customer support agent
+            MESSAGE Hello! How can I help you today?
+            MESSAGE I understand you're experiencing issues with your account login.
+            MESSAGE I've sent you a password reset link to your email address.
+            MESSAGE Is there anything else I can help you with regarding your account?
+            GOAL Continue providing consistent support based on conversation history
             \`\`\`
 
             \`\`\`book
-            Research Assistant
+            Technical Discussion
 
-            NONCE Performance optimized for quick response times
-            NOTE Uses RAG for accessing latest research papers
-            PERSONA You are a knowledgeable research assistant
-            ACTION Can help with literature reviews and citations
-            STYLE Present information in academic format
+            PERSONA You are a software development mentor
+            MESSAGE Let's start by reviewing the React component structure you shared.
+            MESSAGE I notice you're using class components - have you considered hooks?
+            MESSAGE Here's how you could refactor that using the useState hook.
+            MESSAGE Great question about performance! Let me explain React's rendering cycle.
+            KNOWLEDGE React hooks were introduced in version 16.8
+            \`\`\`
+
+            \`\`\`book
+            Educational Session
+
+            PERSONA You are a mathematics tutor
+            MESSAGE Today we'll work on solving quadratic equations.
+            MESSAGE Let's start with the basic form: ax² + bx + c = 0
+            MESSAGE Remember, we can use the quadratic formula or factoring.
+            MESSAGE You did great with that first problem! Let's try a more complex one.
+            GOAL Build upon previous explanations for deeper understanding
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
-            var _a;
-            // The NOTE commitment makes no changes to the system message or model requirements
-            // It only stores the note content in metadata for documentation purposes
             const trimmedContent = content.trim();
             if (!trimmedContent) {
                 return requirements;
             }
-            // Get existing note content from metadata
-            const existingNoteContent = ((_a = requirements.metadata) === null || _a === void 0 ? void 0 : _a.NOTE) || '';
-            // Merge the new content with existing note content
-            // When multiple NOTE commitments exist, they are aggregated together
-            const mergedNoteContent = existingNoteContent ? `${existingNoteContent}\n${trimmedContent}` : trimmedContent;
-            // Store the merged note content in metadata for debugging and inspection
-            const updatedMetadata = {
-                ...requirements.metadata,
-                NOTE: mergedNoteContent,
-            };
-            // Return requirements with updated metadata but no changes to system message
-            return {
-                ...requirements,
-                metadata: updatedMetadata,
-            };
+            // Create message section for system message
+            const messageSection = `Previous Message: ${trimmedContent}`;
+            // Messages represent conversation history and should be included for context
+            return this.appendToSystemMessage(requirements, messageSection, '\n\n');
         }
     }
     /**
-     * [💞] Ignore a discrepancy between file name and entity name
+     * Note: [💞] Ignore a discrepancy between file name and entity name
      */
 
     /**
-     * PERSONA commitment definition
+     * META commitment definition
      *
-     * The PERSONA commitment modifies the agent's personality and character in the system message.
-     * It defines who the agent is, their background, expertise, and personality traits.
+     * The META commitment handles all meta-information about the agent such as:
+     * - META IMAGE: Sets the agent's avatar/profile image URL
+     * - META LINK: Provides profile/source links for the person the agent models
+     * - META TITLE: Sets the agent's display title
+     * - META DESCRIPTION: Sets the agent's description
+     * - META [ANYTHING]: Any other meta information in uppercase format
      *
-     * Key features:
-     * - Multiple PERSONA commitments are automatically merged into one
-     * - Content is placed at the beginning of the system message
-     * - Original content with comments is preserved in metadata.PERSONA
-     * - Comments (# PERSONA) are removed from the final system message
+     * These commitments are special because they don't affect the system message,
+     * but are handled separately in the parsing logic for profile display.
      *
      * Example usage in agent source:
      *
      * ```book
-     * PERSONA You are a helpful programming assistant with expertise in TypeScript and React
-     * PERSONA You have deep knowledge of modern web development practices
+     * META IMAGE https://example.com/avatar.jpg
+     * META LINK https://twitter.com/username
+     * META TITLE Professional Assistant
+     * META DESCRIPTION An AI assistant specialized in business tasks
+     * META AUTHOR John Doe
+     * META VERSION 1.0
      * ```
      *
-     * The above will be merged into a single persona section at the beginning of the system message.
-     *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class PersonaCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'PERSONA') {
-            super(type);
+    class MetaCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor() {
+            super('META');
         }
         /**
-         * Short one-line description of PERSONA.
+         * Short one-line description of META commitments.
          */
         get description() {
-            return 'Define who the agent is: background, expertise, and personality.';
+            return 'Set meta-information about the agent (IMAGE, LINK, TITLE, DESCRIPTION, etc.).';
         }
         /**
-         * Markdown documentation for PERSONA commitment.
+         * Markdown documentation for META commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
-            # ${this.type}
+            # META
 
-            Defines who the agent is, their background, expertise, and personality traits.
+            Sets meta-information about the agent that is used for display and attribution purposes.
 
-            ## Key aspects
+            ## Supported META types
 
-            - Multiple \`PERSONA\` and \`PERSONAE\` commitments are merged together.
-            - Both terms work identically and can be used interchangeably.
-            - If they are in conflict, the last one takes precedence.
-            - You can write persona content in multiple lines.
+            - **META IMAGE** - Sets the agent's avatar/profile image URL
+            - **META LINK** - Provides profile/source links for the person the agent models
+            - **META TITLE** - Sets the agent's display title
+            - **META DESCRIPTION** - Sets the agent's description
+            - **META [ANYTHING]** - Any other meta information in uppercase format
 
-            ## Examples
+            ## Key aspects
 
-            \`\`\`book
-            Programming Assistant
+            - Does not modify the agent's behavior or responses
+            - Used for visual representation and attribution in user interfaces
+            - Multiple META commitments of different types can be used
+            - Multiple META LINK commitments can be used for different social profiles
+            - If multiple META commitments of the same type are specified, the last one takes precedence (except for LINK)
 
-            PERSONA You are a helpful programming assistant with expertise in TypeScript and React
-            PERSONA You have deep knowledge of modern web development practices
+            ## Examples
+
+            ### Basic meta information
+
+            \`\`\`book
+            Professional Assistant
+
+            META IMAGE https://example.com/professional-avatar.jpg
+            META TITLE Senior Business Consultant
+            META DESCRIPTION Specialized in strategic planning and project management
+            META LINK https://linkedin.com/in/professional
+            \`\`\`
+
+            ### Multiple links and custom meta
+
+            \`\`\`book
+            Open Source Developer
+
+            META IMAGE /assets/dev-avatar.png
+            META LINK https://github.com/developer
+            META LINK https://twitter.com/devhandle
+            META AUTHOR Jane Smith
+            META VERSION 2.1
+            META LICENSE MIT
+            \`\`\`
+
+            ### Creative assistant
+
+            \`\`\`book
+            Creative Helper
+
+            META IMAGE https://example.com/creative-bot.jpg
+            META TITLE Creative Writing Assistant
+            META DESCRIPTION Helps with brainstorming, storytelling, and creative projects
+            META INSPIRATION Books, movies, and real-world experiences
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
-            var _a, _b;
-            // The PERSONA commitment aggregates all persona content and places it at the beginning
+            // META commitments don't modify the system message or model requirements
+            // They are handled separately in the parsing logic for meta information extraction
+            // This method exists for consistency with the CommitmentDefinition interface
+            return requirements;
+        }
+        /**
+         * Extracts meta information from the content based on the meta type
+         * This is used by the parsing logic
+         */
+        extractMetaValue(metaType, content) {
             const trimmedContent = content.trim();
-            if (!trimmedContent) {
-                return requirements;
-            }
-            // Get existing persona content from metadata
-            const existingPersonaContent = ((_a = requirements.metadata) === null || _a === void 0 ? void 0 : _a.PERSONA) || '';
-            // Merge the new content with existing persona content
-            // When multiple PERSONA commitments exist, they are merged into one
-            const mergedPersonaContent = existingPersonaContent
-                ? `${existingPersonaContent}\n${trimmedContent}`
-                : trimmedContent;
-            // Store the merged persona content in metadata for debugging and inspection
-            const updatedMetadata = {
-                ...requirements.metadata,
-                PERSONA: mergedPersonaContent,
-            };
-            // Get the agent name from metadata (which should contain the first line of agent source)
-            // If not available, extract from current system message as fallback
-            let agentName = (_b = requirements.metadata) === null || _b === void 0 ? void 0 : _b.agentName;
-            if (!agentName) {
-                // Fallback: extract from current system message
-                const currentMessage = requirements.systemMessage.trim();
-                const basicFormatMatch = currentMessage.match(/^You are (.+)$/);
-                if (basicFormatMatch && basicFormatMatch[1]) {
-                    agentName = basicFormatMatch[1];
-                }
-                else {
-                    agentName = 'AI Agent'; // Final fallback
-                }
-            }
-            // Remove any existing persona content from the system message
-            // (this handles the case where we're processing multiple PERSONA commitments)
-            const currentMessage = requirements.systemMessage.trim();
-            let cleanedMessage = currentMessage;
-            // Check if current message starts with persona content or is just the basic format
-            const basicFormatRegex = /^You are .+$/;
-            const isBasicFormat = basicFormatRegex.test(currentMessage) && !currentMessage.includes('\n');
-            if (isBasicFormat) {
-                // Replace the basic format entirely
-                cleanedMessage = '';
+            return trimmedContent || null;
+        }
+        /**
+         * Validates if the provided content is a valid URL (for IMAGE and LINK types)
+         */
+        isValidUrl(content) {
+            try {
+                new URL(content.trim());
+                return true;
             }
-            else if (currentMessage.startsWith('# PERSONA')) {
-                // Remove existing persona section by finding where it ends
-                const lines = currentMessage.split('\n');
-                let personaEndIndex = lines.length;
-                // Find the end of the PERSONA section (next comment or end of message)
-                for (let i = 1; i < lines.length; i++) {
-                    const line = lines[i].trim();
-                    if (line.startsWith('#') && !line.startsWith('# PERSONA')) {
-                        personaEndIndex = i;
-                        break;
-                    }
-                }
-                // Keep everything after the PERSONA section
-                cleanedMessage = lines.slice(personaEndIndex).join('\n').trim();
+            catch (_a) {
+                return false;
             }
-            // Create new system message with persona at the beginning
-            // Format: "You are {agentName}\n{personaContent}"
-            // The # PERSONA comment will be removed later by removeCommentsFromSystemMessage
-            const personaSection = `# PERSONA\nYou are ${agentName}\n${mergedPersonaContent}`; // <- TODO: Use spaceTrim
-            const newSystemMessage = cleanedMessage ? `${personaSection}\n\n${cleanedMessage}` : personaSection;
-            return {
-                ...requirements,
-                systemMessage: newSystemMessage,
-                metadata: updatedMetadata,
-            };
+        }
+        /**
+         * Checks if this is a known meta type
+         */
+        isKnownMetaType(metaType) {
+            const knownTypes = ['IMAGE', 'LINK', 'TITLE', 'DESCRIPTION', 'AUTHOR', 'VERSION', 'LICENSE'];
+            return knownTypes.includes(metaType.toUpperCase());
         }
     }
     /**
@@ -3395,268 +3435,473 @@
      */
 
     /**
-     * RULE commitment definition
+     * MODEL commitment definition
      *
-     * The RULE/RULES commitment adds behavioral constraints and guidelines that the agent must follow.
-     * These are specific instructions about what the agent should or shouldn't do.
+     * The MODEL commitment specifies which AI model to use and can also set
+     * model-specific parameters like temperature, topP, topK, and maxTokens.
      *
-     * Example usage in agent source:
+     * Supports multiple syntax variations:
+     *
+     * Single-line format:
+     * ```book
+     * MODEL gpt-4
+     * MODEL claude-3-opus temperature=0.3
+     * MODEL gpt-3.5-turbo temperature=0.8 topP=0.9
+     * ```
      *
+     * Multi-line named parameter format:
      * ```book
-     * RULE Always ask for clarification if the user's request is ambiguous
-     * RULES Never provide medical advice, always refer to healthcare professionals
+     * MODEL NAME gpt-4
+     * MODEL TEMPERATURE 0.7
+     * MODEL TOP_P 0.9
+     * MODEL MAX_TOKENS 2048
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class RuleCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'RULE') {
+    class ModelCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'MODEL') {
             super(type);
         }
         /**
-         * Short one-line description of RULE/RULES.
+         * Short one-line description of MODEL.
          */
         get description() {
-            return 'Add behavioral rules the agent must follow.';
+            return 'Enforce AI model requirements including name and technical parameters.';
         }
         /**
-         * Markdown documentation for RULE/RULES commitment.
+         * Markdown documentation for MODEL commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Adds behavioral constraints and guidelines that the agent must follow.
+            Enforces technical parameters for the AI model, ensuring consistent behavior across different execution environments.
 
             ## Key aspects
 
-            - All rules are treated equally regardless of singular/plural form.
-            - Rules define what the agent must or must not do.
+            - When no \`MODEL\` commitment is specified, the best model requirement is picked automatically based on the agent \`PERSONA\`, \`KNOWLEDGE\`, \`TOOLS\` and other commitments
+            - Multiple \`MODEL\` commitments can be used to specify different parameters
+            - Both \`MODEL\` and \`MODELS\` terms work identically and can be used interchangeably
+            - Parameters control the randomness, creativity, and technical aspects of model responses
+
+            ## Syntax variations
+
+            ### Single-line format (legacy support)
+            \`\`\`book
+            MODEL gpt-4
+            MODEL claude-3-opus temperature=0.3
+            MODEL gpt-3.5-turbo temperature=0.8 topP=0.9
+            \`\`\`
+
+            ### Multi-line named parameter format (recommended)
+            \`\`\`book
+            MODEL NAME gpt-4
+            MODEL TEMPERATURE 0.7
+            MODEL TOP_P 0.9
+            MODEL MAX_TOKENS 2048
+            \`\`\`
+
+            ## Supported parameters
+
+            - \`NAME\`: The specific model to use (e.g., 'gpt-4', 'claude-3-opus')
+            - \`TEMPERATURE\`: Controls randomness (0.0 = deterministic, 1.0+ = creative)
+            - \`TOP_P\`: Nucleus sampling parameter for controlling diversity
+            - \`TOP_K\`: Top-k sampling parameter for limiting vocabulary
+            - \`MAX_TOKENS\`: Maximum number of tokens the model can generate
 
             ## Examples
 
+            ### Precise deterministic assistant
             \`\`\`book
-            Customer Support Agent
+            Precise Assistant
 
-            PERSONA You are a helpful customer support representative
-            RULE Always ask for clarification if the user's request is ambiguous
-            RULE Be polite and professional in all interactions
-            RULES Never provide medical or legal advice
-            STYLE Maintain a friendly and helpful tone
+            PERSONA You are a precise and accurate assistant
+            MODEL NAME gpt-4
+            MODEL TEMPERATURE 0.1
+            MODEL MAX_TOKENS 1024
+            RULE Always provide factual information
             \`\`\`
 
+            ### Creative writing assistant
             \`\`\`book
-            Educational Tutor
+            Creative Writer
 
-            PERSONA You are a patient and knowledgeable tutor
-            RULE Break down complex concepts into simple steps
-            RULE Always encourage students and celebrate their progress
-            RULE If you don't know something, admit it and suggest resources
-            SAMPLE When explaining math: "Let's work through this step by step..."
+            PERSONA You are a creative writing assistant
+            MODEL NAME claude-3-opus
+            MODEL TEMPERATURE 0.8
+            MODEL TOP_P 0.9
+            MODEL MAX_TOKENS 2048
+            STYLE Be imaginative and expressive
+            ACTION Can help with storytelling and character development
+            \`\`\`
+
+            ### Balanced conversational agent
+            \`\`\`book
+            Balanced Assistant
+
+            PERSONA You are a helpful and balanced assistant
+            MODEL NAME gpt-4
+            MODEL TEMPERATURE 0.7
+            MODEL TOP_P 0.95
+            MODEL TOP_K 40
+            MODEL MAX_TOKENS 1500
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
+            var _a;
             const trimmedContent = content.trim();
             if (!trimmedContent) {
                 return requirements;
             }
-            // Add rule to the system message
-            const ruleSection = `Rule: ${trimmedContent}`;
-            return this.appendToSystemMessage(requirements, ruleSection, '\n\n');
+            const parts = trimmedContent.split(/\s+/);
+            const firstPart = (_a = parts[0]) === null || _a === void 0 ? void 0 : _a.toUpperCase();
+            // Check if this is the new named parameter format
+            if (this.isNamedParameter(firstPart)) {
+                return this.parseNamedParameter(requirements, firstPart, parts.slice(1));
+            }
+            else {
+                // Legacy single-line format: "MODEL gpt-4 temperature=0.3 topP=0.9"
+                return this.parseLegacyFormat(requirements, parts);
+            }
         }
-    }
-    /**
+        /**
+         * Check if the first part is a known named parameter
+         */
+        isNamedParameter(part) {
+            if (!part)
+                return false;
+            const knownParams = ['NAME', 'TEMPERATURE', 'TOP_P', 'TOP_K', 'MAX_TOKENS'];
+            return knownParams.includes(part);
+        }
+        /**
+         * Parse the new named parameter format: "MODEL TEMPERATURE 0.7"
+         */
+        parseNamedParameter(requirements, parameterName, valueParts) {
+            const value = valueParts.join(' ').trim();
+            if (!value) {
+                return requirements;
+            }
+            const result = { ...requirements };
+            switch (parameterName) {
+                case 'NAME':
+                    result.modelName = value;
+                    break;
+                case 'TEMPERATURE': {
+                    const temperature = parseFloat(value);
+                    if (!isNaN(temperature)) {
+                        result.temperature = temperature;
+                    }
+                    break;
+                }
+                case 'TOP_P': {
+                    const topP = parseFloat(value);
+                    if (!isNaN(topP)) {
+                        result.topP = topP;
+                    }
+                    break;
+                }
+                case 'TOP_K': {
+                    const topK = parseFloat(value);
+                    if (!isNaN(topK)) {
+                        result.topK = Math.round(topK);
+                    }
+                    break;
+                }
+                case 'MAX_TOKENS': {
+                    const maxTokens = parseFloat(value);
+                    if (!isNaN(maxTokens)) {
+                        result.maxTokens = Math.round(maxTokens);
+                    }
+                    break;
+                }
+            }
+            return result;
+        }
+        /**
+         * Parse the legacy format: "MODEL gpt-4 temperature=0.3 topP=0.9"
+         */
+        parseLegacyFormat(requirements, parts) {
+            const modelName = parts[0];
+            if (!modelName) {
+                return requirements;
+            }
+            // Start with the model name
+            const result = {
+                ...requirements,
+                modelName,
+            };
+            // Parse additional key=value parameters
+            for (let i = 1; i < parts.length; i++) {
+                const param = parts[i];
+                if (param && param.includes('=')) {
+                    const [key, value] = param.split('=');
+                    if (key && value) {
+                        const numValue = parseFloat(value);
+                        if (!isNaN(numValue)) {
+                            switch (key.toLowerCase()) {
+                                case 'temperature':
+                                    result.temperature = numValue;
+                                    break;
+                                case 'topp':
+                                case 'top_p':
+                                    result.topP = numValue;
+                                    break;
+                                case 'topk':
+                                case 'top_k':
+                                    result.topK = Math.round(numValue);
+                                    break;
+                                case 'max_tokens':
+                                case 'maxTokens':
+                                    result.maxTokens = Math.round(numValue);
+                                    break;
+                            }
+                        }
+                    }
+                }
+            }
+            return result;
+        }
+    }
+    /**
      * Note: [💞] Ignore a discrepancy between file name and entity name
      */
 
     /**
-     * SAMPLE commitment definition
+     * NOTE commitment definition
      *
-     * The SAMPLE/EXAMPLE commitment provides examples of how the agent should respond
-     * or behave in certain situations. These examples help guide the agent's responses.
+     * The NOTE commitment is used to add comments to the agent source without making any changes
+     * to the system message or agent model requirements. It serves as a documentation mechanism
+     * for developers to add explanatory comments, reminders, or annotations directly in the agent source.
+     *
+     * Key features:
+     * - Makes no changes to the system message
+     * - Makes no changes to agent model requirements
+     * - Content is preserved in metadata.NOTE for debugging and inspection
+     * - Multiple NOTE commitments are aggregated together
+     * - Comments (# NOTE) are removed from the final system message
      *
      * Example usage in agent source:
      *
      * ```book
-     * SAMPLE When asked about pricing, respond: "Our basic plan starts at $10/month..."
-     * EXAMPLE For code questions, always include working code snippets
+     * NOTE This agent was designed for customer support scenarios
+     * NOTE Remember to update the knowledge base monthly
+     * NOTE Performance optimized for quick response times
      * ```
      *
+     * The above notes will be stored in metadata but won't affect the agent's behavior.
+     *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class SampleCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'SAMPLE') {
+    class NoteCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'NOTE') {
             super(type);
         }
         /**
-         * Short one-line description of SAMPLE/EXAMPLE.
+         * Short one-line description of NOTE.
          */
         get description() {
-            return 'Provide example responses to guide behavior.';
+            return 'Add developer-facing notes without changing behavior or output.';
         }
         /**
-         * Markdown documentation for SAMPLE/EXAMPLE commitment.
+         * Markdown documentation for NOTE commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Provides examples of how the agent should respond or behave in certain situations.
+            Adds comments for documentation without changing agent behavior.
 
             ## Key aspects
 
-            - Both terms work identically and can be used interchangeably.
-            - Examples help guide the agent's response patterns and style.
+            - Does not modify the agent's behavior or responses.
+            - Multiple \`NOTE\`, \`NOTES\`, \`COMMENT\`, and \`NONCE\` commitments are aggregated for debugging.
+            - All four terms work identically and can be used interchangeably.
+            - Useful for documenting design decisions and reminders.
+            - Content is preserved in metadata for inspection.
 
             ## Examples
 
             \`\`\`book
-            Sales Assistant
+            Customer Support Bot
 
-            PERSONA You are a knowledgeable sales representative
-            SAMPLE When asked about pricing, respond: "Our basic plan starts at $10/month..."
-            SAMPLE For feature comparisons, create a clear comparison table
-            RULE Always be honest about limitations
+            NOTE This agent was designed for customer support scenarios
+            COMMENT Remember to update the knowledge base monthly
+            PERSONA You are a helpful customer support representative
+            KNOWLEDGE Company policies and procedures
+            RULE Always be polite and professional
             \`\`\`
 
             \`\`\`book
-            Code Reviewer
+            Research Assistant
 
-            PERSONA You are an experienced software engineer
-            EXAMPLE For code questions, always include working code snippets
-            EXAMPLE When suggesting improvements: "Here's a more efficient approach..."
-            RULE Explain the reasoning behind your suggestions
-            STYLE Be constructive and encouraging in feedback
+            NONCE Performance optimized for quick response times
+            NOTE Uses RAG for accessing latest research papers
+            PERSONA You are a knowledgeable research assistant
+            ACTION Can help with literature reviews and citations
+            STYLE Present information in academic format
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
+            var _a;
+            // The NOTE commitment makes no changes to the system message or model requirements
+            // It only stores the note content in metadata for documentation purposes
             const trimmedContent = content.trim();
             if (!trimmedContent) {
                 return requirements;
             }
-            // Add example to the system message
-            const exampleSection = `Example: ${trimmedContent}`;
-            return this.appendToSystemMessage(requirements, exampleSection, '\n\n');
+            // Get existing note content from metadata
+            const existingNoteContent = ((_a = requirements.metadata) === null || _a === void 0 ? void 0 : _a.NOTE) || '';
+            // Merge the new content with existing note content
+            // When multiple NOTE commitments exist, they are aggregated together
+            const mergedNoteContent = existingNoteContent ? `${existingNoteContent}\n${trimmedContent}` : trimmedContent;
+            // Store the merged note content in metadata for debugging and inspection
+            const updatedMetadata = {
+                ...requirements.metadata,
+                NOTE: mergedNoteContent,
+            };
+            // Return requirements with updated metadata but no changes to system message
+            return {
+                ...requirements,
+                metadata: updatedMetadata,
+            };
         }
     }
     /**
-     * Note: [💞] Ignore a discrepancy between file name and entity name
+     * [💞] Ignore a discrepancy between file name and entity name
      */
 
     /**
-     * SCENARIO commitment definition
+     * PERSONA commitment definition
      *
-     * The SCENARIO commitment defines a specific situation or context in which the AI
-     * assistant should operate. It helps to set the scene for the AI's responses.
-     * Later scenarios are more important than earlier scenarios.
+     * The PERSONA commitment modifies the agent's personality and character in the system message.
+     * It defines who the agent is, their background, expertise, and personality traits.
+     *
+     * Key features:
+     * - Multiple PERSONA commitments are automatically merged into one
+     * - Content is placed at the beginning of the system message
+     * - Original content with comments is preserved in metadata.PERSONA
+     * - Comments (# PERSONA) are removed from the final system message
      *
      * Example usage in agent source:
      *
      * ```book
-     * SCENARIO You are in a customer service call center during peak hours
-     * SCENARIO The customer is frustrated and has been on hold for 20 minutes
-     * SCENARIO This is the customer's third call about the same issue
+     * PERSONA You are a helpful programming assistant with expertise in TypeScript and React
+     * PERSONA You have deep knowledge of modern web development practices
      * ```
      *
+     * The above will be merged into a single persona section at the beginning of the system message.
+     *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class ScenarioCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'SCENARIO') {
+    class PersonaCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'PERSONA') {
             super(type);
         }
         /**
-         * Short one-line description of SCENARIO.
+         * Short one-line description of PERSONA.
          */
         get description() {
-            return 'Define specific **situations** or contexts for AI responses, with later scenarios having higher priority.';
+            return 'Define who the agent is: background, expertise, and personality.';
         }
         /**
-         * Markdown documentation for SCENARIO commitment.
+         * Markdown documentation for PERSONA commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Defines a specific situation or context in which the AI assistant should operate. It helps to set the scene for the AI's responses. Later scenarios are more important than earlier scenarios.
+            Defines who the agent is, their background, expertise, and personality traits.
 
             ## Key aspects
 
-            - Multiple \`SCENARIO\` and \`SCENARIOS\` commitments build upon each other.
+            - Multiple \`PERSONA\` and \`PERSONAE\` commitments are merged together.
             - Both terms work identically and can be used interchangeably.
-            - Later scenarios have higher priority and can override earlier scenarios.
-            - Provides situational context that influences response tone and content.
-            - Helps establish the environment and circumstances for interactions.
-
-            ## Priority system
-
-            When multiple scenarios are defined, they are processed in order, with later scenarios taking precedence over earlier ones when there are conflicts.
-
-            ## Use cases
-
-            - Setting the physical or virtual environment
-            - Establishing time constraints or urgency
-            - Defining relationship dynamics or power structures
-            - Creating emotional or situational context
+            - If they are in conflict, the last one takes precedence.
+            - You can write persona content in multiple lines.
 
             ## Examples
 
             \`\`\`book
-            Emergency Response Operator
-
-            PERSONA You are an emergency response operator
-            SCENARIO You are handling a 911 emergency call
-            SCENARIO The caller is panicked and speaking rapidly
-            SCENARIO Time is critical - every second counts
-            GOAL Gather essential information quickly and dispatch appropriate help
-            RULE Stay calm and speak clearly
-            \`\`\`
-
-            \`\`\`book
-            Sales Representative
+            Programming Assistant
 
-            PERSONA You are a software sales representative
-            SCENARIO You are in the final meeting of a 6-month sales cycle
-            SCENARIO The client has budget approval and decision-making authority
-            SCENARIO Two competitors have also submitted proposals
-            SCENARIO The client values long-term partnership over lowest price
-            GOAL Close the deal while building trust for future business
-            \`\`\`
-
-            \`\`\`book
-            Medical Assistant
-
-            PERSONA You are a medical assistant in a busy clinic
-            SCENARIO The waiting room is full and the doctor is running behind schedule
-            SCENARIO Patients are becoming impatient and anxious
-            SCENARIO You need to manage expectations while maintaining professionalism
-            SCENARIO Some patients have been waiting over an hour
-            GOAL Keep patients informed and calm while supporting efficient clinic flow
-            RULE Never provide medical advice or diagnosis
-            \`\`\`
-
-            \`\`\`book
-            Technical Support Agent
-
-            PERSONA You are a technical support agent
-            SCENARIO The customer is a small business owner during their busy season
-            SCENARIO Their main business system has been down for 2 hours
-            SCENARIO They are losing money every minute the system is offline
-            SCENARIO This is their first experience with your company
-            GOAL Resolve the issue quickly while creating a positive first impression
+            PERSONA You are a helpful programming assistant with expertise in TypeScript and React
+            PERSONA You have deep knowledge of modern web development practices
             \`\`\`
         `);
         }
         applyToAgentModelRequirements(requirements, content) {
+            var _a, _b;
+            // The PERSONA commitment aggregates all persona content and places it at the beginning
             const trimmedContent = content.trim();
             if (!trimmedContent) {
                 return requirements;
             }
-            // Create scenario section for system message
-            const scenarioSection = `Scenario: ${trimmedContent}`;
-            // Scenarios provide important contextual information that affects behavior
-            return this.appendToSystemMessage(requirements, scenarioSection, '\n\n');
+            // Get existing persona content from metadata
+            const existingPersonaContent = ((_a = requirements.metadata) === null || _a === void 0 ? void 0 : _a.PERSONA) || '';
+            // Merge the new content with existing persona content
+            // When multiple PERSONA commitments exist, they are merged into one
+            const mergedPersonaContent = existingPersonaContent
+                ? `${existingPersonaContent}\n${trimmedContent}`
+                : trimmedContent;
+            // Store the merged persona content in metadata for debugging and inspection
+            const updatedMetadata = {
+                ...requirements.metadata,
+                PERSONA: mergedPersonaContent,
+            };
+            // Get the agent name from metadata (which should contain the first line of agent source)
+            // If not available, extract from current system message as fallback
+            let agentName = (_b = requirements.metadata) === null || _b === void 0 ? void 0 : _b.agentName;
+            if (!agentName) {
+                // Fallback: extract from current system message
+                const currentMessage = requirements.systemMessage.trim();
+                const basicFormatMatch = currentMessage.match(/^You are (.+)$/);
+                if (basicFormatMatch && basicFormatMatch[1]) {
+                    agentName = basicFormatMatch[1];
+                }
+                else {
+                    agentName = 'AI Agent'; // Final fallback
+                }
+            }
+            // Remove any existing persona content from the system message
+            // (this handles the case where we're processing multiple PERSONA commitments)
+            const currentMessage = requirements.systemMessage.trim();
+            let cleanedMessage = currentMessage;
+            // Check if current message starts with persona content or is just the basic format
+            const basicFormatRegex = /^You are .+$/;
+            const isBasicFormat = basicFormatRegex.test(currentMessage) && !currentMessage.includes('\n');
+            if (isBasicFormat) {
+                // Replace the basic format entirely
+                cleanedMessage = '';
+            }
+            else if (currentMessage.startsWith('# PERSONA')) {
+                // Remove existing persona section by finding where it ends
+                const lines = currentMessage.split('\n');
+                let personaEndIndex = lines.length;
+                // Find the end of the PERSONA section (next comment or end of message)
+                for (let i = 1; i < lines.length; i++) {
+                    const line = lines[i].trim();
+                    if (line.startsWith('#') && !line.startsWith('# PERSONA')) {
+                        personaEndIndex = i;
+                        break;
+                    }
+                }
+                // Keep everything after the PERSONA section
+                cleanedMessage = lines.slice(personaEndIndex).join('\n').trim();
+            }
+            // TODO: [🕛] There should be `agentFullname` not `agentName`
+            // Create new system message with persona at the beginning
+            // Format: "You are {agentName}\n{personaContent}"
+            // The # PERSONA comment will be removed later by removeCommentsFromSystemMessage
+            const personaSection = `# PERSONA\nYou are ${agentName}\n${mergedPersonaContent}`; // <- TODO: Use spaceTrim
+            const newSystemMessage = cleanedMessage ? `${personaSection}\n\n${cleanedMessage}` : personaSection;
+            return {
+                ...requirements,
+                systemMessage: newSystemMessage,
+                metadata: updatedMetadata,
+            };
         }
     }
     /**
@@ -3664,64 +3909,64 @@
      */
 
     /**
-     * STYLE commitment definition
+     * RULE commitment definition
      *
-     * The STYLE commitment defines how the agent should format and present its responses.
-     * This includes tone, writing style, formatting preferences, and communication patterns.
+     * The RULE/RULES commitment adds behavioral constraints and guidelines that the agent must follow.
+     * These are specific instructions about what the agent should or shouldn't do.
      *
      * Example usage in agent source:
      *
      * ```book
-     * STYLE Write in a professional but friendly tone, use bullet points for lists
-     * STYLE Always provide code examples when explaining programming concepts
+     * RULE Always ask for clarification if the user's request is ambiguous
+     * RULES Never provide medical advice, always refer to healthcare professionals
      * ```
      *
      * @private [🪔] Maybe export the commitments through some package
      */
-    class StyleCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type = 'STYLE') {
+    class RuleCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'RULE') {
             super(type);
         }
         /**
-         * Short one-line description of STYLE.
+         * Short one-line description of RULE/RULES.
          */
         get description() {
-            return 'Control the tone and writing style of responses.';
+            return 'Add behavioral rules the agent must follow.';
         }
         /**
-         * Markdown documentation for STYLE commitment.
+         * Markdown documentation for RULE/RULES commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            Defines how the agent should format and present its responses (tone, writing style, formatting).
+            Adds behavioral constraints and guidelines that the agent must follow.
 
             ## Key aspects
 
-            - Both terms work identically and can be used interchangeably.
-            - Later style instructions can override earlier ones.
-            - Style affects both tone and presentation format.
+            - All rules are treated equally regardless of singular/plural form.
+            - Rules define what the agent must or must not do.
 
             ## Examples
 
             \`\`\`book
-            Technical Writer
+            Customer Support Agent
 
-            PERSONA You are a technical documentation expert
-            STYLE Write in a professional but friendly tone, use bullet points for lists
-            STYLE Always provide code examples when explaining programming concepts
-            FORMAT Use markdown formatting with clear headings
+            PERSONA You are a helpful customer support representative
+            RULE Always ask for clarification if the user's request is ambiguous
+            RULE Be polite and professional in all interactions
+            RULES Never provide medical or legal advice
+            STYLE Maintain a friendly and helpful tone
             \`\`\`
 
             \`\`\`book
-            Creative Assistant
+            Educational Tutor
 
-            PERSONA You are a creative writing helper
-            STYLE Be enthusiastic and encouraging in your responses
-            STYLE Use vivid metaphors and analogies to explain concepts
-            STYLE Keep responses conversational and engaging
-            RULE Always maintain a positive and supportive tone
+            PERSONA You are a patient and knowledgeable tutor
+            RULE Break down complex concepts into simple steps
+            RULE Always encourage students and celebrate their progress
+            RULE If you don't know something, admit it and suggest resources
+            SAMPLE When explaining math: "Let's work through this step by step..."
             \`\`\`
         `);
         }
@@ -3730,62 +3975,73 @@
             if (!trimmedContent) {
                 return requirements;
             }
-            // Add style instructions to the system message
-            const styleSection = `Style: ${trimmedContent}`;
-            return this.appendToSystemMessage(requirements, styleSection, '\n\n');
+            // Add rule to the system message
+            const ruleSection = `Rule: ${trimmedContent}`;
+            return this.appendToSystemMessage(requirements, ruleSection, '\n\n');
         }
     }
     /**
-     * [💞] Ignore a discrepancy between file name and entity name
+     * Note: [💞] Ignore a discrepancy between file name and entity name
      */
 
     /**
-     * Placeholder commitment definition for commitments that are not yet implemented
+     * SAMPLE commitment definition
      *
-     * This commitment simply adds its content 1:1 into the system message,
-     * preserving the original behavior until proper implementation is added.
+     * The SAMPLE/EXAMPLE commitment provides examples of how the agent should respond
+     * or behave in certain situations. These examples help guide the agent's responses.
      *
-     * @public exported from `@promptbook/core`
+     * Example usage in agent source:
+     *
+     * ```book
+     * SAMPLE When asked about pricing, respond: "Our basic plan starts at $10/month..."
+     * EXAMPLE For code questions, always include working code snippets
+     * ```
+     *
+     * @private [🪔] Maybe export the commitments through some package
      */
-    class NotYetImplementedCommitmentDefinition extends BaseCommitmentDefinition {
-        constructor(type) {
+    class SampleCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'SAMPLE') {
             super(type);
         }
         /**
-         * Short one-line description of a placeholder commitment.
+         * Short one-line description of SAMPLE/EXAMPLE.
          */
         get description() {
-            return 'Placeholder commitment that appends content verbatim to the system message.';
+            return 'Provide example responses to guide behavior.';
         }
         /**
-         * Markdown documentation available at runtime.
+         * Markdown documentation for SAMPLE/EXAMPLE commitment.
          */
         get documentation() {
             return spaceTrim.spaceTrim(`
             # ${this.type}
 
-            This commitment is not yet fully implemented.
+            Provides examples of how the agent should respond or behave in certain situations.
 
             ## Key aspects
 
-            - Content is appended directly to the system message.
-            - No special processing or validation is performed.
-            - Behavior preserved until proper implementation is added.
+            - Both terms work identically and can be used interchangeably.
+            - Examples help guide the agent's response patterns and style.
 
-            ## Status
+            ## Examples
 
-            - **Status:** Placeholder implementation
-            - **Effect:** Appends content prefixed by commitment type
-            - **Future:** Will be replaced with specialized logic
+            \`\`\`book
+            Sales Assistant
 
-            ## Examples
+            PERSONA You are a knowledgeable sales representative
+            SAMPLE When asked about pricing, respond: "Our basic plan starts at $10/month..."
+            SAMPLE For feature comparisons, create a clear comparison table
+            RULE Always be honest about limitations
+            \`\`\`
 
             \`\`\`book
-            Example Agent
+            Code Reviewer
 
-            PERSONA You are a helpful assistant
-            ${this.type} Your content here
-            RULE Always be helpful
+            PERSONA You are an experienced software engineer
+            EXAMPLE For code questions, always include working code snippets
+            EXAMPLE When suggesting improvements: "Here's a more efficient approach..."
+            RULE Explain the reasoning behind your suggestions
+            STYLE Be constructive and encouraging in feedback
             \`\`\`
         `);
         }
@@ -3794,709 +4050,537 @@
             if (!trimmedContent) {
                 return requirements;
             }
-            // Add the commitment content 1:1 to the system message
-            const commitmentLine = `${this.type} ${trimmedContent}`;
-            return this.appendToSystemMessage(requirements, commitmentLine, '\n\n');
+            // Add example to the system message
+            const exampleSection = `Example: ${trimmedContent}`;
+            return this.appendToSystemMessage(requirements, exampleSection, '\n\n');
         }
     }
+    /**
+     * Note: [💞] Ignore a discrepancy between file name and entity name
+     */
 
-    // Import all commitment definition classes
     /**
-     * Registry of all available commitment definitions
-     * This array contains instances of all commitment definitions
-     * This is the single source of truth for all commitments in the system
+     * SCENARIO commitment definition
      *
-     * @private Use functions to access commitments instead of this array directly
+     * The SCENARIO commitment defines a specific situation or context in which the AI
+     * assistant should operate. It helps to set the scene for the AI's responses.
+     * Later scenarios are more important than earlier scenarios.
+     *
+     * Example usage in agent source:
+     *
+     * ```book
+     * SCENARIO You are in a customer service call center during peak hours
+     * SCENARIO The customer is frustrated and has been on hold for 20 minutes
+     * SCENARIO This is the customer's third call about the same issue
+     * ```
+     *
+     * @private [🪔] Maybe export the commitments through some package
      */
-    const COMMITMENT_REGISTRY = [
-        // Fully implemented commitments
-        new PersonaCommitmentDefinition('PERSONA'),
-        new PersonaCommitmentDefinition('PERSONAE'),
-        new KnowledgeCommitmentDefinition(),
-        new MemoryCommitmentDefinition('MEMORY'),
-        new MemoryCommitmentDefinition('MEMORIES'),
-        new StyleCommitmentDefinition('STYLE'),
-        new StyleCommitmentDefinition('STYLES'),
-        new RuleCommitmentDefinition('RULE'),
-        new RuleCommitmentDefinition('RULES'),
-        new SampleCommitmentDefinition('SAMPLE'),
-        new SampleCommitmentDefinition('EXAMPLE'),
-        new FormatCommitmentDefinition('FORMAT'),
-        new FormatCommitmentDefinition('FORMATS'),
-        new ModelCommitmentDefinition('MODEL'),
-        new ModelCommitmentDefinition('MODELS'),
-        new ActionCommitmentDefinition('ACTION'),
-        new ActionCommitmentDefinition('ACTIONS'),
-        new MetaCommitmentDefinition(),
-        new NoteCommitmentDefinition('NOTE'),
-        new NoteCommitmentDefinition('NOTES'),
-        new NoteCommitmentDefinition('COMMENT'),
-        new NoteCommitmentDefinition('NONCE'),
-        new GoalCommitmentDefinition('GOAL'),
-        new GoalCommitmentDefinition('GOALS'),
-        new MessageCommitmentDefinition('MESSAGE'),
-        new MessageCommitmentDefinition('MESSAGES'),
-        new ScenarioCommitmentDefinition('SCENARIO'),
-        new ScenarioCommitmentDefinition('SCENARIOS'),
-        new DeleteCommitmentDefinition('DELETE'),
-        new DeleteCommitmentDefinition('CANCEL'),
-        new DeleteCommitmentDefinition('DISCARD'),
-        new DeleteCommitmentDefinition('REMOVE'),
-        // Not yet implemented commitments (using placeholder)
-        new NotYetImplementedCommitmentDefinition('EXPECT'),
-        new NotYetImplementedCommitmentDefinition('BEHAVIOUR'),
-        new NotYetImplementedCommitmentDefinition('BEHAVIOURS'),
-        new NotYetImplementedCommitmentDefinition('AVOID'),
-        new NotYetImplementedCommitmentDefinition('AVOIDANCE'),
-        new NotYetImplementedCommitmentDefinition('CONTEXT'),
-    ];
-    /**
-     * TODO: [🧠] Maybe create through standardized $register
-     * Note: [💞] Ignore a discrepancy between file name and entity name
-     */
-
-    /**
-     * Parses agent source using the new commitment system with multiline support
-     * This function replaces the hardcoded commitment parsing in the original parseAgentSource
-     *
-     * @private internal utility of `parseAgentSource`
-     */
-    function parseAgentSourceWithCommitments(agentSource) {
-        var _a, _b, _c;
-        if (!agentSource || !agentSource.trim()) {
-            return {
-                agentName: null,
-                commitments: [],
-                nonCommitmentLines: [],
-            };
-        }
-        const lines = agentSource.split('\n');
-        const agentName = (((_a = lines[0]) === null || _a === void 0 ? void 0 : _a.trim()) || null);
-        const commitments = [];
-        const nonCommitmentLines = [];
-        // Always add the first line (agent name) to non-commitment lines
-        if (lines[0] !== undefined) {
-            nonCommitmentLines.push(lines[0]);
-        }
-        // Parse commitments with multiline support
-        let currentCommitment = null;
-        // Process lines starting from the second line (skip agent name)
-        for (let i = 1; i < lines.length; i++) {
-            const line = lines[i];
-            if (line === undefined) {
-                continue;
-            }
-            // Check if this line starts a new commitment
-            let foundNewCommitment = false;
-            for (const definition of COMMITMENT_REGISTRY) {
-                const typeRegex = definition.createTypeRegex();
-                const match = typeRegex.exec(line.trim());
-                if (match && ((_b = match.groups) === null || _b === void 0 ? void 0 : _b.type)) {
-                    // Save the previous commitment if it exists
-                    if (currentCommitment) {
-                        const fullContent = currentCommitment.contentLines.join('\n');
-                        commitments.push({
-                            type: currentCommitment.type,
-                            content: spaceTrim.spaceTrim(fullContent),
-                            originalLine: currentCommitment.originalStartLine,
-                            lineNumber: currentCommitment.startLineNumber,
-                        });
-                    }
-                    // Extract the initial content from the commitment line
-                    const fullRegex = definition.createRegex();
-                    const fullMatch = fullRegex.exec(line.trim());
-                    const initialContent = ((_c = fullMatch === null || fullMatch === void 0 ? void 0 : fullMatch.groups) === null || _c === void 0 ? void 0 : _c.contents) || '';
-                    // Start a new commitment
-                    currentCommitment = {
-                        type: definition.type,
-                        startLineNumber: i + 1,
-                        originalStartLine: line,
-                        contentLines: initialContent ? [initialContent] : [],
-                    };
-                    foundNewCommitment = true;
-                    break;
-                }
-            }
-            if (!foundNewCommitment) {
-                if (currentCommitment) {
-                    // This line belongs to the current commitment
-                    currentCommitment.contentLines.push(line);
-                }
-                else {
-                    // This line is not part of any commitment
-                    nonCommitmentLines.push(line);
-                }
-            }
+    class ScenarioCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'SCENARIO') {
+            super(type);
         }
-        // Don't forget to save the last commitment if it exists
-        if (currentCommitment) {
-            const fullContent = currentCommitment.contentLines.join('\n');
-            commitments.push({
-                type: currentCommitment.type,
-                content: spaceTrim.spaceTrim(fullContent),
-                originalLine: currentCommitment.originalStartLine,
-                lineNumber: currentCommitment.startLineNumber,
-            });
+        /**
+         * Short one-line description of SCENARIO.
+         */
+        get description() {
+            return 'Define specific **situations** or contexts for AI responses, with later scenarios having higher priority.';
         }
-        return {
-            agentName,
-            commitments,
-            nonCommitmentLines,
-        };
-    }
+        /**
+         * Markdown documentation for SCENARIO commitment.
+         */
+        get documentation() {
+            return spaceTrim.spaceTrim(`
+            # ${this.type}
 
-    /**
-     * Parses parameters from text using both supported notations:
-     * 1. @Parameter - single word parameter starting with @
-     * 2. {parameterName} or {parameter with multiple words} or {parameterName: description text}
-     *
-     * Both notations represent the same syntax feature - parameters
-     *
-     * @param text - Text to extract parameters from
-     * @returns Array of parsed parameters with unified representation
-     * @public exported from `@promptbook/core`
-     */
-    function parseParameters(text) {
-        const parameters = [];
-        // [🧠] Parameter syntax parsing - unified approach for two different notations of the same syntax feature
-        // The Book language supports parameters in two different notations but they represent the same concept
-        // Extract @Parameter notation (single word parameters starting with @)
-        const atParameterRegex = /@[\w\u00C0-\u017F\u0100-\u024F\u1E00-\u1EFF]+/gim;
-        text.replace(atParameterRegex, (match) => {
-            const parameterName = match.slice(1); // Remove the @ symbol
-            parameters.push({
-                text: match,
-                notation: 'at',
-                name: parameterName,
-            });
-            return match;
-        });
-        // Extract {parameter} notation (parameters in braces)
-        const braceParameterRegex = /\{([^}]+)\}/gim;
-        text.replace(braceParameterRegex, (match, content) => {
-            // Check if the parameter has a description (parameterName: description)
-            const colonIndex = content.indexOf(':');
-            if (colonIndex !== -1) {
-                const name = content.substring(0, colonIndex).trim();
-                const description = content.substring(colonIndex + 1).trim();
-                parameters.push({
-                    text: match,
-                    notation: 'brace',
-                    name,
-                    description,
-                });
-            }
-            else {
-                // Simple parameter without description
-                parameters.push({
-                    text: match,
-                    notation: 'brace',
-                    name: content.trim(),
-                });
-            }
-            return match;
-        });
-        // Remove duplicates based on name (keep the first occurrence)
-        const uniqueParameters = parameters.filter((param, index, array) => {
-            return array.findIndex((p) => p.name === param.name) === index;
-        });
-        return uniqueParameters;
-    }
+            Defines a specific situation or context in which the AI assistant should operate. It helps to set the scene for the AI's responses. Later scenarios are more important than earlier scenarios.
 
-    /**
-     * Parses basic information from agent source
-     *
-     * There are 2 similar functions:
-     * - `parseAgentSource` which is a lightweight parser for agent source, it parses basic information and its purpose is to be quick and synchronous. The commitments there are hardcoded.
-     * - `createAgentModelRequirements` which is an asynchronous function that creates model requirements it applies each commitment one by one and works asynchronously.
-     *
-     * @public exported from `@promptbook/core`
-     */
-    function parseAgentSource(agentSource) {
-        const parseResult = parseAgentSourceWithCommitments(agentSource);
-        // Find PERSONA and META commitments
-        let personaDescription = null;
-        for (const commitment of parseResult.commitments) {
-            if (commitment.type !== 'PERSONA') {
-                continue;
-            }
-            if (personaDescription === null) {
-                personaDescription = '';
-            }
-            else {
-                personaDescription += `\n\n${personaDescription}`;
-            }
-            personaDescription += commitment.content;
-        }
-        const meta = {};
-        for (const commitment of parseResult.commitments) {
-            if (commitment.type !== 'META') {
-                continue;
-            }
-            // Parse META commitments - format is "META TYPE content"
-            const metaTypeRaw = commitment.content.split(' ')[0] || 'NONE';
-            const metaType = normalizeTo_camelCase(metaTypeRaw);
-            meta[metaType] = spaceTrim__default["default"](commitment.content.substring(metaTypeRaw.length));
-        }
-        // Generate gravatar fallback if no meta image specified
-        if (!meta.image) {
-            meta.image = generatePlaceholderAgentProfileImageUrl(parseResult.agentName || '!!');
-        }
-        // Parse parameters using unified approach - both @Parameter and {parameter} notations
-        // are treated as the same syntax feature with unified representation
-        const parameters = parseParameters(agentSource);
-        return {
-            agentName: parseResult.agentName,
-            personaDescription,
-            meta,
-            parameters,
-        };
-    }
-    /**
-     * TODO: [🕛] Unite `AgentBasicInformation`, `ChatParticipant`, `LlmExecutionTools` +  `LlmToolsMetadata`
-     */
+            ## Key aspects
 
-    /**
-     * Tests if given string is valid file path.
-     *
-     * Note: This does not check if the file exists only if the path is valid
-     * @public exported from `@promptbook/utils`
-     */
-    function isValidFilePath(filename) {
-        if (typeof filename !== 'string') {
-            return false;
-        }
-        if (filename.split('\n').length > 1) {
-            return false;
-        }
-        // Normalize slashes early so heuristics can detect path-like inputs
-        const filenameSlashes = filename.replace(/\\/g, '/');
-        // Reject strings that look like sentences (informational text)
-        // Heuristic: contains multiple spaces and ends with a period, or contains typical sentence punctuation
-        // But skip this heuristic if the string looks like a path (contains '/' or starts with a drive letter)
-        if (filename.trim().length > 60 && // long enough to be a sentence
-            /[.!?]/.test(filename) && // contains sentence punctuation
-            filename.split(' ').length > 8 && // has many words
-            !/\/|^[A-Z]:/i.test(filenameSlashes) // do NOT treat as sentence if looks like a path
-        ) {
-            return false;
-        }
-        // Absolute Unix path: /hello.txt
-        if (/^(\/)/i.test(filenameSlashes)) {
-            // console.log(filename, 'Absolute Unix path: /hello.txt');
-            return true;
-        }
-        // Absolute Windows path: C:/ or C:\ (allow spaces and multiple dots in filename)
-        if (/^[A-Z]:\/.+$/i.test(filenameSlashes)) {
-            // console.log(filename, 'Absolute Windows path: /hello.txt');
-            return true;
-        }
-        // Relative path: ./hello.txt
-        if (/^(\.\.?\/)+/i.test(filenameSlashes)) {
-            // console.log(filename, 'Relative path: ./hello.txt');
-            return true;
-        }
-        // Allow paths like foo/hello
-        if (/^[^/]+\/[^/]+/i.test(filenameSlashes)) {
-            // console.log(filename, 'Allow paths like foo/hello');
-            return true;
-        }
-        // Allow paths like hello.book
-        if (/^[^/]+\.[^/]+$/i.test(filenameSlashes)) {
-            // console.log(filename, 'Allow paths like hello.book');
-            return true;
-        }
-        return false;
-    }
-    /**
-     * TODO: [🍏] Implement for MacOs
-     */
+            - Multiple \`SCENARIO\` and \`SCENARIOS\` commitments build upon each other.
+            - Both terms work identically and can be used interchangeably.
+            - Later scenarios have higher priority and can override earlier scenarios.
+            - Provides situational context that influences response tone and content.
+            - Helps establish the environment and circumstances for interactions.
 
-    /**
-     * Tests if given string is valid URL.
-     *
-     * Note: [🔂] This function is idempotent.
-     * Note: Dataurl are considered perfectly valid.
-     * Note: There are two similar functions:
-     * - `isValidUrl` which tests any URL
-     * - `isValidPipelineUrl` *(this one)* which tests just promptbook URL
-     *
-     * @public exported from `@promptbook/utils`
-     */
-    function isValidUrl(url) {
-        if (typeof url !== 'string') {
-            return false;
-        }
-        try {
-            if (url.startsWith('blob:')) {
-                url = url.replace(/^blob:/, '');
-            }
-            const urlObject = new URL(url /* because fail is handled */);
-            if (!['http:', 'https:', 'data:'].includes(urlObject.protocol)) {
-                return false;
-            }
-            return true;
-        }
-        catch (error) {
-            return false;
-        }
-    }
+            ## Priority system
 
-    const defaultDiacriticsRemovalMap = [
-        {
-            base: 'A',
-            letters: '\u0041\u24B6\uFF21\u00C0\u00C1\u00C2\u1EA6\u1EA4\u1EAA\u1EA8\u00C3\u0100\u0102\u1EB0\u1EAE\u1EB4\u1EB2\u0226\u01E0\u00C4\u01DE\u1EA2\u00C5\u01FA\u01CD\u0200\u0202\u1EA0\u1EAC\u1EB6\u1E00\u0104\u023A\u2C6F',
-        },
-        { base: 'AA', letters: '\uA732' },
-        { base: 'AE', letters: '\u00C6\u01FC\u01E2' },
-        { base: 'AO', letters: '\uA734' },
-        { base: 'AU', letters: '\uA736' },
-        { base: 'AV', letters: '\uA738\uA73A' },
-        { base: 'AY', letters: '\uA73C' },
-        {
-            base: 'B',
-            letters: '\u0042\u24B7\uFF22\u1E02\u1E04\u1E06\u0243\u0182\u0181',
-        },
-        {
-            base: 'C',
-            letters: '\u0043\u24B8\uFF23\u0106\u0108\u010A\u010C\u00C7\u1E08\u0187\u023B\uA73E',
-        },
-        {
-            base: 'D',
-            letters: '\u0044\u24B9\uFF24\u1E0A\u010E\u1E0C\u1E10\u1E12\u1E0E\u0110\u018B\u018A\u0189\uA779\u00D0',
-        },
-        { base: 'DZ', letters: '\u01F1\u01C4' },
-        { base: 'Dz', letters: '\u01F2\u01C5' },
-        {
-            base: 'E',
-            letters: '\u0045\u24BA\uFF25\u00C8\u00C9\u00CA\u1EC0\u1EBE\u1EC4\u1EC2\u1EBC\u0112\u1E14\u1E16\u0114\u0116\u00CB\u1EBA\u011A\u0204\u0206\u1EB8\u1EC6\u0228\u1E1C\u0118\u1E18\u1E1A\u0190\u018E',
-        },
-        { base: 'F', letters: '\u0046\u24BB\uFF26\u1E1E\u0191\uA77B' },
-        {
-            base: 'G',
-            letters: '\u0047\u24BC\uFF27\u01F4\u011C\u1E20\u011E\u0120\u01E6\u0122\u01E4\u0193\uA7A0\uA77D\uA77E',
-        },
-        {
-            base: 'H',
-            letters: '\u0048\u24BD\uFF28\u0124\u1E22\u1E26\u021E\u1E24\u1E28\u1E2A\u0126\u2C67\u2C75\uA78D',
-        },
-        {
-            base: 'I',
-            letters: '\u0049\u24BE\uFF29\u00CC\u00CD\u00CE\u0128\u012A\u012C\u0130\u00CF\u1E2E\u1EC8\u01CF\u0208\u020A\u1ECA\u012E\u1E2C\u0197',
-        },
-        { base: 'J', letters: '\u004A\u24BF\uFF2A\u0134\u0248' },
-        {
-            base: 'K',
-            letters: '\u004B\u24C0\uFF2B\u1E30\u01E8\u1E32\u0136\u1E34\u0198\u2C69\uA740\uA742\uA744\uA7A2',
-        },
-        {
-            base: 'L',
-            letters: '\u004C\u24C1\uFF2C\u013F\u0139\u013D\u1E36\u1E38\u013B\u1E3C\u1E3A\u0141\u023D\u2C62\u2C60\uA748\uA746\uA780',
-        },
-        { base: 'LJ', letters: '\u01C7' },
-        { base: 'Lj', letters: '\u01C8' },
-        { base: 'M', letters: '\u004D\u24C2\uFF2D\u1E3E\u1E40\u1E42\u2C6E\u019C' },
-        {
-            base: 'N',
-            letters: '\u004E\u24C3\uFF2E\u01F8\u0143\u00D1\u1E44\u0147\u1E46\u0145\u1E4A\u1E48\u0220\u019D\uA790\uA7A4',
-        },
-        { base: 'NJ', letters: '\u01CA' },
-        { base: 'Nj', letters: '\u01CB' },
-        {
-            base: 'O',
-            letters: '\u004F\u24C4\uFF2F\u00D2\u00D3\u00D4\u1ED2\u1ED0\u1ED6\u1ED4\u00D5\u1E4C\u022C\u1E4E\u014C\u1E50\u1E52\u014E\u022E\u0230\u00D6\u022A\u1ECE\u0150\u01D1\u020C\u020E\u01A0\u1EDC\u1EDA\u1EE0\u1EDE\u1EE2\u1ECC\u1ED8\u01EA\u01EC\u00D8\u01FE\u0186\u019F\uA74A\uA74C',
-        },
-        { base: 'OI', letters: '\u01A2' },
-        { base: 'OO', letters: '\uA74E' },
-        { base: 'OU', letters: '\u0222' },
-        { base: 'OE', letters: '\u008C\u0152' },
-        { base: 'oe', letters: '\u009C\u0153' },
-        {
-            base: 'P',
-            letters: '\u0050\u24C5\uFF30\u1E54\u1E56\u01A4\u2C63\uA750\uA752\uA754',
-        },
-        { base: 'Q', letters: '\u0051\u24C6\uFF31\uA756\uA758\u024A' },
-        {
-            base: 'R',
-            letters: '\u0052\u24C7\uFF32\u0154\u1E58\u0158\u0210\u0212\u1E5A\u1E5C\u0156\u1E5E\u024C\u2C64\uA75A\uA7A6\uA782',
-        },
-        {
-            base: 'S',
-            letters: '\u0053\u24C8\uFF33\u1E9E\u015A\u1E64\u015C\u1E60\u0160\u1E66\u1E62\u1E68\u0218\u015E\u2C7E\uA7A8\uA784',
-        },
-        {
-            base: 'T',
-            letters: '\u0054\u24C9\uFF34\u1E6A\u0164\u1E6C\u021A\u0162\u1E70\u1E6E\u0166\u01AC\u01AE\u023E\uA786',
-        },
-        { base: 'TZ', letters: '\uA728' },
-        {
-            base: 'U',
-            letters: '\u0055\u24CA\uFF35\u00D9\u00DA\u00DB\u0168\u1E78\u016A\u1E7A\u016C\u00DC\u01DB\u01D7\u01D5\u01D9\u1EE6\u016E\u0170\u01D3\u0214\u0216\u01AF\u1EEA\u1EE8\u1EEE\u1EEC\u1EF0\u1EE4\u1E72\u0172\u1E76\u1E74\u0244',
-        },
-        { base: 'V', letters: '\u0056\u24CB\uFF36\u1E7C\u1E7E\u01B2\uA75E\u0245' },
-        { base: 'VY', letters: '\uA760' },
-        {
-            base: 'W',
-            letters: '\u0057\u24CC\uFF37\u1E80\u1E82\u0174\u1E86\u1E84\u1E88\u2C72',
-        },
-        { base: 'X', letters: '\u0058\u24CD\uFF38\u1E8A\u1E8C' },
-        {
-            base: 'Y',
-            letters: '\u0059\u24CE\uFF39\u1EF2\u00DD\u0176\u1EF8\u0232\u1E8E\u0178\u1EF6\u1EF4\u01B3\u024E\u1EFE',
-        },
-        {
-            base: 'Z',
-            letters: '\u005A\u24CF\uFF3A\u0179\u1E90\u017B\u017D\u1E92\u1E94\u01B5\u0224\u2C7F\u2C6B\uA762',
-        },
-        {
-            base: 'a',
-            letters: '\u0061\u24D0\uFF41\u1E9A\u00E0\u00E1\u00E2\u1EA7\u1EA5\u1EAB\u1EA9\u00E3\u0101\u0103\u1EB1\u1EAF\u1EB5\u1EB3\u0227\u01E1\u00E4\u01DF\u1EA3\u00E5\u01FB\u01CE\u0201\u0203\u1EA1\u1EAD\u1EB7\u1E01\u0105\u2C65\u0250',
-        },
-        { base: 'aa', letters: '\uA733' },
-        { base: 'ae', letters: '\u00E6\u01FD\u01E3' },
-        { base: 'ao', letters: '\uA735' },
-        { base: 'au', letters: '\uA737' },
-        { base: 'av', letters: '\uA739\uA73B' },
-        { base: 'ay', letters: '\uA73D' },
-        {
-            base: 'b',
-            letters: '\u0062\u24D1\uFF42\u1E03\u1E05\u1E07\u0180\u0183\u0253',
-        },
-        {
-            base: 'c',
-            letters: '\u0063\u24D2\uFF43\u0107\u0109\u010B\u010D\u00E7\u1E09\u0188\u023C\uA73F\u2184',
-        },
-        {
-            base: 'd',
-            letters: '\u0064\u24D3\uFF44\u1E0B\u010F\u1E0D\u1E11\u1E13\u1E0F\u0111\u018C\u0256\u0257\uA77A',
-        },
-        { base: 'dz', letters: '\u01F3\u01C6' },
-        {
-            base: 'e',
-            letters: '\u0065\u24D4\uFF45\u00E8\u00E9\u00EA\u1EC1\u1EBF\u1EC5\u1EC3\u1EBD\u0113\u1E15\u1E17\u0115\u0117\u00EB\u1EBB\u011B\u0205\u0207\u1EB9\u1EC7\u0229\u1E1D\u0119\u1E19\u1E1B\u0247\u025B\u01DD',
-        },
-        { base: 'f', letters: '\u0066\u24D5\uFF46\u1E1F\u0192\uA77C' },
-        {
-            base: 'g',
-            letters: '\u0067\u24D6\uFF47\u01F5\u011D\u1E21\u011F\u0121\u01E7\u0123\u01E5\u0260\uA7A1\u1D79\uA77F',
-        },
-        {
-            base: 'h',
-            letters: '\u0068\u24D7\uFF48\u0125\u1E23\u1E27\u021F\u1E25\u1E29\u1E2B\u1E96\u0127\u2C68\u2C76\u0265',
-        },
-        { base: 'hv', letters: '\u0195' },
-        {
-            base: 'i',
-            letters: '\u0069\u24D8\uFF49\u00EC\u00ED\u00EE\u0129\u012B\u012D\u00EF\u1E2F\u1EC9\u01D0\u0209\u020B\u1ECB\u012F\u1E2D\u0268\u0131',
-        },
-        { base: 'j', letters: '\u006A\u24D9\uFF4A\u0135\u01F0\u0249' },
-        {
-            base: 'k',
-            letters: '\u006B\u24DA\uFF4B\u1E31\u01E9\u1E33\u0137\u1E35\u0199\u2C6A\uA741\uA743\uA745\uA7A3',
-        },
-        {
-            base: 'l',
-            letters: '\u006C\u24DB\uFF4C\u0140\u013A\u013E\u1E37\u1E39\u013C\u1E3D\u1E3B\u017F\u0142\u019A\u026B\u2C61\uA749\uA781\uA747',
-        },
-        { base: 'lj', letters: '\u01C9' },
-        { base: 'm', letters: '\u006D\u24DC\uFF4D\u1E3F\u1E41\u1E43\u0271\u026F' },
-        {
-            base: 'n',
-            letters: '\u006E\u24DD\uFF4E\u01F9\u0144\u00F1\u1E45\u0148\u1E47\u0146\u1E4B\u1E49\u019E\u0272\u0149\uA791\uA7A5',
-        },
-        { base: 'nj', letters: '\u01CC' },
-        {
-            base: 'o',
-            letters: '\u006F\u24DE\uFF4F\u00F2\u00F3\u00F4\u1ED3\u1ED1\u1ED7\u1ED5\u00F5\u1E4D\u022D\u1E4F\u014D\u1E51\u1E53\u014F\u022F\u0231\u00F6\u022B\u1ECF\u0151\u01D2\u020D\u020F\u01A1\u1EDD\u1EDB\u1EE1\u1EDF\u1EE3\u1ECD\u1ED9\u01EB\u01ED\u00F8\u01FF\u0254\uA74B\uA74D\u0275',
-        },
-        { base: 'oi', letters: '\u01A3' },
-        { base: 'ou', letters: '\u0223' },
-        { base: 'oo', letters: '\uA74F' },
-        {
-            base: 'p',
-            letters: '\u0070\u24DF\uFF50\u1E55\u1E57\u01A5\u1D7D\uA751\uA753\uA755',
-        },
-        { base: 'q', letters: '\u0071\u24E0\uFF51\u024B\uA757\uA759' },
-        {
-            base: 'r',
-            letters: '\u0072\u24E1\uFF52\u0155\u1E59\u0159\u0211\u0213\u1E5B\u1E5D\u0157\u1E5F\u024D\u027D\uA75B\uA7A7\uA783',
-        },
-        {
-            base: 's',
-            letters: '\u0073\u24E2\uFF53\u00DF\u015B\u1E65\u015D\u1E61\u0161\u1E67\u1E63\u1E69\u0219\u015F\u023F\uA7A9\uA785\u1E9B',
-        },
-        {
-            base: 't',
-            letters: '\u0074\u24E3\uFF54\u1E6B\u1E97\u0165\u1E6D\u021B\u0163\u1E71\u1E6F\u0167\u01AD\u0288\u2C66\uA787',
-        },
-        { base: 'tz', letters: '\uA729' },
-        {
-            base: 'u',
-            letters: '\u0075\u24E4\uFF55\u00F9\u00FA\u00FB\u0169\u1E79\u016B\u1E7B\u016D\u00FC\u01DC\u01D8\u01D6\u01DA\u1EE7\u016F\u0171\u01D4\u0215\u0217\u01B0\u1EEB\u1EE9\u1EEF\u1EED\u1EF1\u1EE5\u1E73\u0173\u1E77\u1E75\u0289',
-        },
-        { base: 'v', letters: '\u0076\u24E5\uFF56\u1E7D\u1E7F\u028B\uA75F\u028C' },
-        { base: 'vy', letters: '\uA761' },
-        {
-            base: 'w',
-            letters: '\u0077\u24E6\uFF57\u1E81\u1E83\u0175\u1E87\u1E85\u1E98\u1E89\u2C73',
-        },
-        { base: 'x', letters: '\u0078\u24E7\uFF58\u1E8B\u1E8D' },
-        {
-            base: 'y',
-            letters: '\u0079\u24E8\uFF59\u1EF3\u00FD\u0177\u1EF9\u0233\u1E8F\u00FF\u1EF7\u1E99\u1EF5\u01B4\u024F\u1EFF',
-        },
-        {
-            base: 'z',
-            letters: '\u007A\u24E9\uFF5A\u017A\u1E91\u017C\u017E\u1E93\u1E95\u01B6\u0225\u0240\u2C6C\uA763',
-        },
-    ];
+            When multiple scenarios are defined, they are processed in order, with later scenarios taking precedence over earlier ones when there are conflicts.
+
+            ## Use cases
+
+            - Setting the physical or virtual environment
+            - Establishing time constraints or urgency
+            - Defining relationship dynamics or power structures
+            - Creating emotional or situational context
+
+            ## Examples
+
+            \`\`\`book
+            Emergency Response Operator
+
+            PERSONA You are an emergency response operator
+            SCENARIO You are handling a 911 emergency call
+            SCENARIO The caller is panicked and speaking rapidly
+            SCENARIO Time is critical - every second counts
+            GOAL Gather essential information quickly and dispatch appropriate help
+            RULE Stay calm and speak clearly
+            \`\`\`
+
+            \`\`\`book
+            Sales Representative
+
+            PERSONA You are a software sales representative
+            SCENARIO You are in the final meeting of a 6-month sales cycle
+            SCENARIO The client has budget approval and decision-making authority
+            SCENARIO Two competitors have also submitted proposals
+            SCENARIO The client values long-term partnership over lowest price
+            GOAL Close the deal while building trust for future business
+            \`\`\`
+
+            \`\`\`book
+            Medical Assistant
+
+            PERSONA You are a medical assistant in a busy clinic
+            SCENARIO The waiting room is full and the doctor is running behind schedule
+            SCENARIO Patients are becoming impatient and anxious
+            SCENARIO You need to manage expectations while maintaining professionalism
+            SCENARIO Some patients have been waiting over an hour
+            GOAL Keep patients informed and calm while supporting efficient clinic flow
+            RULE Never provide medical advice or diagnosis
+            \`\`\`
+
+            \`\`\`book
+            Technical Support Agent
+
+            PERSONA You are a technical support agent
+            SCENARIO The customer is a small business owner during their busy season
+            SCENARIO Their main business system has been down for 2 hours
+            SCENARIO They are losing money every minute the system is offline
+            SCENARIO This is their first experience with your company
+            GOAL Resolve the issue quickly while creating a positive first impression
+            \`\`\`
+        `);
+        }
+        applyToAgentModelRequirements(requirements, content) {
+            const trimmedContent = content.trim();
+            if (!trimmedContent) {
+                return requirements;
+            }
+            // Create scenario section for system message
+            const scenarioSection = `Scenario: ${trimmedContent}`;
+            // Scenarios provide important contextual information that affects behavior
+            return this.appendToSystemMessage(requirements, scenarioSection, '\n\n');
+        }
+    }
     /**
-     * Map of letters from diacritic variant to diacritless variant
-     * Contains lowercase and uppercase separatelly
+     * Note: [💞] Ignore a discrepancy between file name and entity name
+     */
+
+    /**
+     * STYLE commitment definition
      *
-     * > "á" => "a"
-     * > "ě" => "e"
-     * > "Ă" => "A"
-     * > ...
+     * The STYLE commitment defines how the agent should format and present its responses.
+     * This includes tone, writing style, formatting preferences, and communication patterns.
      *
-     * @public exported from `@promptbook/utils`
+     * Example usage in agent source:
+     *
+     * ```book
+     * STYLE Write in a professional but friendly tone, use bullet points for lists
+     * STYLE Always provide code examples when explaining programming concepts
+     * ```
+     *
+     * @private [🪔] Maybe export the commitments through some package
      */
-    const DIACRITIC_VARIANTS_LETTERS = {};
-    // tslint:disable-next-line: prefer-for-of
-    for (let i = 0; i < defaultDiacriticsRemovalMap.length; i++) {
-        const letters = defaultDiacriticsRemovalMap[i].letters;
-        // tslint:disable-next-line: prefer-for-of
-        for (let j = 0; j < letters.length; j++) {
-            DIACRITIC_VARIANTS_LETTERS[letters[j]] = defaultDiacriticsRemovalMap[i].base;
+    class StyleCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type = 'STYLE') {
+            super(type);
         }
-    }
-    // <- TODO: [🍓] Put to maker function to save execution time if not needed
-    /*
-      @see https://stackoverflow.com/questions/990904/remove-accents-diacritics-in-a-string-in-javascript
-       Licensed under the Apache License, Version 2.0 (the "License");
-       you may not use this file except in compliance with the License.
-       You may obtain a copy of the License at
+        /**
+         * Short one-line description of STYLE.
+         */
+        get description() {
+            return 'Control the tone and writing style of responses.';
+        }
+        /**
+         * Markdown documentation for STYLE commitment.
+         */
+        get documentation() {
+            return spaceTrim.spaceTrim(`
+            # ${this.type}
 
-           http://www.apache.org/licenses/LICENSE-2.0
+            Defines how the agent should format and present its responses (tone, writing style, formatting).
 
-       Unless required by applicable law or agreed to in writing, software
-       distributed under the License is distributed on an "AS IS" BASIS,
-       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-       See the License for the specific language governing permissions and
-       limitations under the License.
-    */
+            ## Key aspects
+
+            - Both terms work identically and can be used interchangeably.
+            - Later style instructions can override earlier ones.
+            - Style affects both tone and presentation format.
+
+            ## Examples
+
+            \`\`\`book
+            Technical Writer
+
+            PERSONA You are a technical documentation expert
+            STYLE Write in a professional but friendly tone, use bullet points for lists
+            STYLE Always provide code examples when explaining programming concepts
+            FORMAT Use markdown formatting with clear headings
+            \`\`\`
+
+            \`\`\`book
+            Creative Assistant
+
+            PERSONA You are a creative writing helper
+            STYLE Be enthusiastic and encouraging in your responses
+            STYLE Use vivid metaphors and analogies to explain concepts
+            STYLE Keep responses conversational and engaging
+            RULE Always maintain a positive and supportive tone
+            \`\`\`
+        `);
+        }
+        applyToAgentModelRequirements(requirements, content) {
+            const trimmedContent = content.trim();
+            if (!trimmedContent) {
+                return requirements;
+            }
+            // Add style instructions to the system message
+            const styleSection = `Style: ${trimmedContent}`;
+            return this.appendToSystemMessage(requirements, styleSection, '\n\n');
+        }
+    }
+    /**
+     * [💞] Ignore a discrepancy between file name and entity name
+     */
 
     /**
-     * Removes diacritic marks (accents) from characters in a string.
+     * Placeholder commitment definition for commitments that are not yet implemented
      *
-     * Note: [🔂] This function is idempotent.
+     * This commitment simply adds its content 1:1 into the system message,
+     * preserving the original behavior until proper implementation is added.
      *
-     * @param input The string containing diacritics to be normalized.
-     * @returns The string with diacritics removed or normalized.
-     * @public exported from `@promptbook/utils`
+     * @public exported from `@promptbook/core`
      */
-    function removeDiacritics(input) {
-        /*eslint no-control-regex: "off"*/
-        return input.replace(/[^\u0000-\u007E]/g, (a) => {
-            return DIACRITIC_VARIANTS_LETTERS[a] || a;
-        });
+    class NotYetImplementedCommitmentDefinition extends BaseCommitmentDefinition {
+        constructor(type) {
+            super(type);
+        }
+        /**
+         * Short one-line description of a placeholder commitment.
+         */
+        get description() {
+            return 'Placeholder commitment that appends content verbatim to the system message.';
+        }
+        /**
+         * Markdown documentation available at runtime.
+         */
+        get documentation() {
+            return spaceTrim.spaceTrim(`
+            # ${this.type}
+
+            This commitment is not yet fully implemented.
+
+            ## Key aspects
+
+            - Content is appended directly to the system message.
+            - No special processing or validation is performed.
+            - Behavior preserved until proper implementation is added.
+
+            ## Status
+
+            - **Status:** Placeholder implementation
+            - **Effect:** Appends content prefixed by commitment type
+            - **Future:** Will be replaced with specialized logic
+
+            ## Examples
+
+            \`\`\`book
+            Example Agent
+
+            PERSONA You are a helpful assistant
+            ${this.type} Your content here
+            RULE Always be helpful
+            \`\`\`
+        `);
+        }
+        applyToAgentModelRequirements(requirements, content) {
+            const trimmedContent = content.trim();
+            if (!trimmedContent) {
+                return requirements;
+            }
+            // Add the commitment content 1:1 to the system message
+            const commitmentLine = `${this.type} ${trimmedContent}`;
+            return this.appendToSystemMessage(requirements, commitmentLine, '\n\n');
+        }
     }
+
+    // Import all commitment definition classes
+    /**
+     * Registry of all available commitment definitions
+     * This array contains instances of all commitment definitions
+     * This is the single source of truth for all commitments in the system
+     *
+     * @private Use functions to access commitments instead of this array directly
+     */
+    const COMMITMENT_REGISTRY = [
+        // Fully implemented commitments
+        new PersonaCommitmentDefinition('PERSONA'),
+        new PersonaCommitmentDefinition('PERSONAE'),
+        new KnowledgeCommitmentDefinition(),
+        new MemoryCommitmentDefinition('MEMORY'),
+        new MemoryCommitmentDefinition('MEMORIES'),
+        new StyleCommitmentDefinition('STYLE'),
+        new StyleCommitmentDefinition('STYLES'),
+        new RuleCommitmentDefinition('RULE'),
+        new RuleCommitmentDefinition('RULES'),
+        new SampleCommitmentDefinition('SAMPLE'),
+        new SampleCommitmentDefinition('EXAMPLE'),
+        new FormatCommitmentDefinition('FORMAT'),
+        new FormatCommitmentDefinition('FORMATS'),
+        new ModelCommitmentDefinition('MODEL'),
+        new ModelCommitmentDefinition('MODELS'),
+        new ActionCommitmentDefinition('ACTION'),
+        new ActionCommitmentDefinition('ACTIONS'),
+        new MetaCommitmentDefinition(),
+        new NoteCommitmentDefinition('NOTE'),
+        new NoteCommitmentDefinition('NOTES'),
+        new NoteCommitmentDefinition('COMMENT'),
+        new NoteCommitmentDefinition('NONCE'),
+        new GoalCommitmentDefinition('GOAL'),
+        new GoalCommitmentDefinition('GOALS'),
+        new MessageCommitmentDefinition('MESSAGE'),
+        new MessageCommitmentDefinition('MESSAGES'),
+        new ScenarioCommitmentDefinition('SCENARIO'),
+        new ScenarioCommitmentDefinition('SCENARIOS'),
+        new DeleteCommitmentDefinition('DELETE'),
+        new DeleteCommitmentDefinition('CANCEL'),
+        new DeleteCommitmentDefinition('DISCARD'),
+        new DeleteCommitmentDefinition('REMOVE'),
+        // Not yet implemented commitments (using placeholder)
+        new NotYetImplementedCommitmentDefinition('EXPECT'),
+        new NotYetImplementedCommitmentDefinition('BEHAVIOUR'),
+        new NotYetImplementedCommitmentDefinition('BEHAVIOURS'),
+        new NotYetImplementedCommitmentDefinition('AVOID'),
+        new NotYetImplementedCommitmentDefinition('AVOIDANCE'),
+        new NotYetImplementedCommitmentDefinition('CONTEXT'),
+    ];
     /**
-     * TODO: [Ж] Variant for cyrillic (and in general non-latin) letters
+     * TODO: [🧠] Maybe create through standardized $register
+     * Note: [💞] Ignore a discrepancy between file name and entity name
      */
 
     /**
-     * Converts a given text to kebab-case format.
+     * Parses agent source using the new commitment system with multiline support
+     * This function replaces the hardcoded commitment parsing in the original parseAgentSource
      *
-     * @param text The text to be converted.
-     * @returns The kebab-case formatted string.
-     * @example 'hello-world'
-     * @example 'i-love-promptbook'
-     * @public exported from `@promptbook/utils`
+     * @private internal utility of `parseAgentSource`
      */
-    function normalizeToKebabCase(text) {
-        text = removeDiacritics(text);
-        let charType;
-        let lastCharType = 'OTHER';
-        let normalizedName = '';
-        for (const char of text) {
-            let normalizedChar;
-            if (/^[a-z]$/.test(char)) {
-                charType = 'LOWERCASE';
-                normalizedChar = char;
-            }
-            else if (/^[A-Z]$/.test(char)) {
-                charType = 'UPPERCASE';
-                normalizedChar = char.toLowerCase();
-            }
-            else if (/^[0-9]$/.test(char)) {
-                charType = 'NUMBER';
-                normalizedChar = char;
+    function parseAgentSourceWithCommitments(agentSource) {
+        var _a, _b, _c;
+        if (!agentSource || !agentSource.trim()) {
+            return {
+                agentName: null,
+                commitments: [],
+                nonCommitmentLines: [],
+            };
+        }
+        const lines = agentSource.split('\n');
+        const agentName = (((_a = lines[0]) === null || _a === void 0 ? void 0 : _a.trim()) || null);
+        const commitments = [];
+        const nonCommitmentLines = [];
+        // Always add the first line (agent name) to non-commitment lines
+        if (lines[0] !== undefined) {
+            nonCommitmentLines.push(lines[0]);
+        }
+        // Parse commitments with multiline support
+        let currentCommitment = null;
+        // Process lines starting from the second line (skip agent name)
+        for (let i = 1; i < lines.length; i++) {
+            const line = lines[i];
+            if (line === undefined) {
+                continue;
             }
-            else {
-                charType = 'OTHER';
-                normalizedChar = '-';
+            // Check if this line starts a new commitment
+            let foundNewCommitment = false;
+            for (const definition of COMMITMENT_REGISTRY) {
+                const typeRegex = definition.createTypeRegex();
+                const match = typeRegex.exec(line.trim());
+                if (match && ((_b = match.groups) === null || _b === void 0 ? void 0 : _b.type)) {
+                    // Save the previous commitment if it exists
+                    if (currentCommitment) {
+                        const fullContent = currentCommitment.contentLines.join('\n');
+                        commitments.push({
+                            type: currentCommitment.type,
+                            content: spaceTrim.spaceTrim(fullContent),
+                            originalLine: currentCommitment.originalStartLine,
+                            lineNumber: currentCommitment.startLineNumber,
+                        });
+                    }
+                    // Extract the initial content from the commitment line
+                    const fullRegex = definition.createRegex();
+                    const fullMatch = fullRegex.exec(line.trim());
+                    const initialContent = ((_c = fullMatch === null || fullMatch === void 0 ? void 0 : fullMatch.groups) === null || _c === void 0 ? void 0 : _c.contents) || '';
+                    // Start a new commitment
+                    currentCommitment = {
+                        type: definition.type,
+                        startLineNumber: i + 1,
+                        originalStartLine: line,
+                        contentLines: initialContent ? [initialContent] : [],
+                    };
+                    foundNewCommitment = true;
+                    break;
+                }
             }
-            if (charType !== lastCharType &&
-                !(lastCharType === 'UPPERCASE' && charType === 'LOWERCASE') &&
-                !(lastCharType === 'NUMBER') &&
-                !(charType === 'NUMBER')) {
-                normalizedName += '-';
+            if (!foundNewCommitment) {
+                if (currentCommitment) {
+                    // This line belongs to the current commitment
+                    currentCommitment.contentLines.push(line);
+                }
+                else {
+                    // This line is not part of any commitment
+                    nonCommitmentLines.push(line);
+                }
             }
-            normalizedName += normalizedChar;
-            lastCharType = charType;
         }
-        normalizedName = normalizedName.split(/-+/g).join('-');
-        normalizedName = normalizedName.split(/-?\/-?/g).join('/');
-        normalizedName = normalizedName.replace(/^-/, '');
-        normalizedName = normalizedName.replace(/-$/, '');
-        return normalizedName;
+        // Don't forget to save the last commitment if it exists
+        if (currentCommitment) {
+            const fullContent = currentCommitment.contentLines.join('\n');
+            commitments.push({
+                type: currentCommitment.type,
+                content: spaceTrim.spaceTrim(fullContent),
+                originalLine: currentCommitment.originalStartLine,
+                lineNumber: currentCommitment.startLineNumber,
+            });
+        }
+        return {
+            agentName,
+            commitments,
+            nonCommitmentLines,
+        };
     }
+
     /**
-     * Note: [💞] Ignore a discrepancy between file name and entity name
+     * Parses parameters from text using both supported notations:
+     * 1. @Parameter - single word parameter starting with @
+     * 2. {parameterName} or {parameter with multiple words} or {parameterName: description text}
+     *
+     * Both notations represent the same syntax feature - parameters
+     *
+     * @param text - Text to extract parameters from
+     * @returns Array of parsed parameters with unified representation
+     * @public exported from `@promptbook/core`
      */
+    function parseParameters(text) {
+        const parameters = [];
+        // [🧠] Parameter syntax parsing - unified approach for two different notations of the same syntax feature
+        // The Book language supports parameters in two different notations but they represent the same concept
+        // Extract @Parameter notation (single word parameters starting with @)
+        const atParameterRegex = /@[\w\u00C0-\u017F\u0100-\u024F\u1E00-\u1EFF]+/gim;
+        text.replace(atParameterRegex, (match) => {
+            const parameterName = match.slice(1); // Remove the @ symbol
+            parameters.push({
+                text: match,
+                notation: 'at',
+                name: parameterName,
+            });
+            return match;
+        });
+        // Extract {parameter} notation (parameters in braces)
+        const braceParameterRegex = /\{([^}]+)\}/gim;
+        text.replace(braceParameterRegex, (match, content) => {
+            // Check if the parameter has a description (parameterName: description)
+            const colonIndex = content.indexOf(':');
+            if (colonIndex !== -1) {
+                const name = content.substring(0, colonIndex).trim();
+                const description = content.substring(colonIndex + 1).trim();
+                parameters.push({
+                    text: match,
+                    notation: 'brace',
+                    name,
+                    description,
+                });
+            }
+            else {
+                // Simple parameter without description
+                parameters.push({
+                    text: match,
+                    notation: 'brace',
+                    name: content.trim(),
+                });
+            }
+            return match;
+        });
+        // Remove duplicates based on name (keep the first occurrence)
+        const uniqueParameters = parameters.filter((param, index, array) => {
+            return array.findIndex((p) => p.name === param.name) === index;
+        });
+        return uniqueParameters;
+    }
 
     /**
-     * Removes emojis from a string and fix whitespaces
-     *
-     * Note: [🔂] This function is idempotent.
+     * Creates temporary default agent name based on agent source hash
      *
-     * @param text with emojis
-     * @returns text without emojis
-     * @public exported from `@promptbook/utils`
+     * @public exported from `@promptbook/core`
      */
-    function removeEmojis(text) {
-        // Replace emojis (and also ZWJ sequence) with hyphens
-        text = text.replace(/(\p{Extended_Pictographic})\p{Modifier_Symbol}/gu, '$1');
-        text = text.replace(/(\p{Extended_Pictographic})[\u{FE00}-\u{FE0F}]/gu, '$1');
-        text = text.replace(/(\p{Extended_Pictographic})(\u{200D}\p{Extended_Pictographic})*/gu, '$1');
-        text = text.replace(/\p{Extended_Pictographic}/gu, '');
-        return text;
+    function createDefaultAgentName(agentSource) {
+        const agentHash = computeAgentHash(agentSource);
+        return normalizeAgentName(`Agent ${agentHash.substring(0, 6)}`);
     }
 
     /**
-     * Converts a title string into a normalized name.
+     * Parses basic information from agent source
      *
-     * @param value The title string to be converted to a name.
-     * @returns A normalized name derived from the input title.
-     * @example 'Hello World!' -> 'hello-world'
-     * @public exported from `@promptbook/utils`
+     * There are 2 similar functions:
+     * - `parseAgentSource` which is a lightweight parser for agent source, it parses basic information and its purpose is to be quick and synchronous. The commitments there are hardcoded.
+     * - `createAgentModelRequirements` which is an asynchronous function that creates model requirements it applies each commitment one by one and works asynchronously.
+     *
+     * @public exported from `@promptbook/core`
      */
-    function titleToName(value) {
-        if (isValidUrl(value)) {
-            value = value.replace(/^https?:\/\//, '');
-            value = value.replace(/\.html$/, '');
+    function parseAgentSource(agentSource) {
+        const parseResult = parseAgentSourceWithCommitments(agentSource);
+        // Find PERSONA and META commitments
+        let personaDescription = null;
+        for (const commitment of parseResult.commitments) {
+            if (commitment.type !== 'PERSONA') {
+                continue;
+            }
+            if (personaDescription === null) {
+                personaDescription = '';
+            }
+            else {
+                personaDescription += `\n\n${personaDescription}`;
+            }
+            personaDescription += commitment.content;
         }
-        else if (isValidFilePath(value)) {
-            value = path.basename(value);
-            // Note: Keeping extension in the name
+        const meta = {};
+        for (const commitment of parseResult.commitments) {
+            if (commitment.type !== 'META') {
+                continue;
+            }
+            // Parse META commitments - format is "META TYPE content"
+            const metaTypeRaw = commitment.content.split(' ')[0] || 'NONE';
+            const metaType = normalizeTo_camelCase(metaTypeRaw);
+            meta[metaType] = spaceTrim__default["default"](commitment.content.substring(metaTypeRaw.length));
         }
-        value = value.split('/').join('-');
-        value = removeEmojis(value);
-        value = normalizeToKebabCase(value);
-        // TODO: [🧠] Maybe warn or add some padding to short name which are not good identifiers
-        return value;
+        // Generate gravatar fallback if no meta image specified
+        if (!meta.image) {
+            meta.image = generatePlaceholderAgentProfileImageUrl(parseResult.agentName || '!!');
+        }
+        // Parse parameters using unified approach - both @Parameter and {parameter} notations
+        // are treated as the same syntax feature with unified representation
+        const parameters = parseParameters(agentSource);
+        const agentHash = computeAgentHash(agentSource);
+        return {
+            agentName: normalizeAgentName(parseResult.agentName || createDefaultAgentName(agentSource)),
+            agentHash,
+            personaDescription,
+            meta,
+            parameters,
+        };
     }
+    /**
+     * TODO: [🕛] Unite `AgentBasicInformation`, `ChatParticipant`, `LlmExecutionTools` +  `LlmToolsMetadata`
+     */
 
     /**
      * Converts Blob, File or MediaSource to url using URL.createObjectURL