fraim-framework 2.0.95 → 2.0.97

package/dist/src/core/utils/stub-generator.js

@@ -1,7 +1,5 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.generateWorkflowStub = generateWorkflowStub;
-exports.parseRegistryWorkflow = parseRegistryWorkflow;
 exports.generateJobStub = generateJobStub;
 exports.generateSkillStub = generateSkillStub;
 exports.generateRuleStub = generateRuleStub;
@@ -46,54 +44,22 @@ function extractLeadParagraph(content) {
     return paragraphLines.join(' ').trim();
 }
 /**
- * Generates a lightweight markdown stub for a workflow.
- * These stubs are committed to the user's repo for discoverability.
+ * Coaching stubs are discoverability artifacts and should be resolved with get_fraim_file.
  */
-function generateWorkflowStub(workflowName, workflowPath, intent, principles) {
-    return `${STUB_MARKER}
-# FRAIM Workflow: ${workflowName}
-
-> [!IMPORTANT]
-> This is a **FRAIM-managed workflow stub**. 
-> To load the full context (rules, templates, and execution steps), ask your AI agent to:
-> \`@fraim get_fraim_workflow("${workflowName}")\`
->
-> DO NOT EXECUTE.
+function generateJobStub(jobName, _jobPath, intent, outcome, steps) {
+    return `${STUB_MARKER}
+# FRAIM Job: ${jobName}
 
 ## Intent
 ${intent}
-`;
-}
-/**
- * Parses a workflow file from the registry to extract its intent and principles for the stub.
- */
-function parseRegistryWorkflow(content) {
-    // Case-insensitive, handles different line endings, optional following header or EOF
-    const intentMatch = content.match(/##\s*intent\s+([\s\S]*?)(?=\n##|$)/i);
-    const principlesMatch = content.match(/##\s*principles\s+([\s\S]*?)(?=\n##|$)/i);
-    const intent = intentMatch ? intentMatch[1].trim() : 'No intent defined.';
-    const principles = principlesMatch
-        ? principlesMatch[1].trim().split(/\r?\n/).map(l => l.replace(/^- /, '').trim()).filter(l => l)
-        : [];
-    return { intent, principles };
-}
-/**
- * Coaching stubs are discoverability artifacts and should be resolved with get_fraim_file.
- */
-function generateJobStub(jobName, _jobPath, intent, outcome, steps) {
-    return `${STUB_MARKER}
-# FRAIM Job: ${jobName}
-
-## Intent
-${intent}
-
-## Outcome
-${outcome}
-
-## Steps
-${steps}
-
----
+
+## Outcome
+${outcome}
+
+## Steps
+${steps}
+
+---
 
 > [!IMPORTANT]
 > **For AI Agents:** Do NOT attempt to execute this job based on the Intent/Outcome above.
@@ -110,41 +76,41 @@ ${steps}
  * Generates a lightweight markdown stub for a skill.
  */
 function generateSkillStub(skillName, skillPath, skillInput, skillOutput) {
-    return `${STUB_MARKER}
-# FRAIM Skill: ${skillName}
-
-## Skill Input
-${skillInput}
-
-## Skill Output
-${skillOutput}
-
----
-
-> [!IMPORTANT]
-> **For AI Agents:** This is a discoverability stub for the skill.
-> All execution details must be fetched from MCP before use.
-> To retrieve the complete skill instructions, call:
-> \`get_fraim_file({ path: "skills/${skillPath}" })\`
+    return `${STUB_MARKER}
+# FRAIM Skill: ${skillName}
+
+## Skill Input
+${skillInput}
+
+## Skill Output
+${skillOutput}
+
+---
+
+> [!IMPORTANT]
+> **For AI Agents:** This is a discoverability stub for the skill.
+> All execution details must be fetched from MCP before use.
+> To retrieve the complete skill instructions, call:
+> \`get_fraim_file({ path: "skills/${skillPath}" })\`
 `;
 }
 /**
  * Generates a lightweight markdown stub for a rule.
  */
 function generateRuleStub(ruleName, rulePath, intent) {
-    return `${STUB_MARKER}
-# FRAIM Rule: ${ruleName}
-
-## Intent
-${intent}
-
----
-
-> [!IMPORTANT]
-> **For AI Agents:** This is a discoverability stub for the rule.
-> All rule details must be fetched from MCP before use.
-> To retrieve the complete rule instructions, call:
-> \`get_fraim_file({ path: "rules/${rulePath}" })\`
+    return `${STUB_MARKER}
+# FRAIM Rule: ${ruleName}
+
+## Intent
+${intent}
+
+---
+
+> [!IMPORTANT]
+> **For AI Agents:** This is a discoverability stub for the rule.
+> All rule details must be fetched from MCP before use.
+> To retrieve the complete rule instructions, call:
+> \`get_fraim_file({ path: "rules/${rulePath}" })\`
 `;
 }
 /**

package/dist/src/core/utils/workflow-parser.js

@@ -5,13 +5,15 @@ const fs_1 = require("fs");
 const path_1 = require("path");
 class WorkflowParser {
     static extractMetadataBlock(content) {
-        const frontmatterMatch = content.match(/^---\r?\n([\s\S]+?)\r?\n---/);
+        // Allow leading comments and whitespace before frontmatter
+        const frontmatterMatch = content.match(/^[\s\S]*?---\r?\n([\s\S]+?)\r?\n---/);
         if (frontmatterMatch) {
             try {
+                const startIndex = frontmatterMatch.index || 0;
                 return {
                     state: 'valid',
                     metadata: JSON.parse(frontmatterMatch[1]),
-                    bodyStartIndex: frontmatterMatch[0].length
+                    bodyStartIndex: startIndex + frontmatterMatch[0].length
                 };
             }
             catch {
@@ -107,7 +109,7 @@ class WorkflowParser {
             overview = contentAfterMetadata;
         }
         const phases = new Map();
-        const phaseSections = restOfContent.split(/^##\s+Phase:\s+/m);
+        const phaseSections = restOfContent.split(/^##\s+Phase:\s*/im);
         if (!metadata.phases) {
             metadata.phases = {};
         }

package/dist/src/local-mcp-server/learning-context-builder.js

@@ -0,0 +1,229 @@
+"use strict";
+/**
+ * LearningContextBuilder builds the learning context section for a user.
+ *
+ * Lives in the local proxy so learning files are always read from the correct
+ * workspace root on the user's machine.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.buildLearningContextSection = buildLearningContextSection;
+const fs_1 = require("fs");
+const path_1 = require("path");
+const project_fraim_paths_1 = require("../core/utils/project-fraim-paths");
+const LEARNINGS_REL = (0, project_fraim_paths_1.getWorkspaceFraimDisplayPath)('personalized-employee/learnings').replace(/\/$/, '');
+const DEFAULT_THRESHOLD = 3.0;
+function getLearningsBase(workspaceRoot) {
+    return (0, path_1.join)(workspaceRoot, LEARNINGS_REL);
+}
+function getScoreThreshold(workspaceRoot) {
+    try {
+        const configPath = (0, project_fraim_paths_1.getWorkspaceConfigPath)(workspaceRoot);
+        if ((0, fs_1.existsSync)(configPath)) {
+            const config = JSON.parse((0, fs_1.readFileSync)(configPath, 'utf8'));
+            const t = config?.learning?.scoreThreshold;
+            if (typeof t === 'number' && t > 0)
+                return t;
+        }
+    }
+    catch {
+        // Fall through to default.
+    }
+    return DEFAULT_THRESHOLD;
+}
+function computeEffectiveScore(severity, lastSeenDate, recurrences, fileType) {
+    const baseScore = severity === 'P-HIGH' ? 8 : severity === 'P-MED' ? 5 : 3;
+    const halfLife = fileType === 'mistake-patterns' ? 90 : 180;
+    let daysSinceLastSeen = 0;
+    try {
+        const lastSeen = new Date(lastSeenDate);
+        const now = new Date();
+        daysSinceLastSeen = Math.max(0, (now.getTime() - lastSeen.getTime()) / (1000 * 60 * 60 * 24));
+    }
+    catch {
+        // Treat as fresh.
+    }
+    const decay = Math.pow(0.5, daysSinceLastSeen / halfLife);
+    const recurrenceBoost = Math.log2(Math.max(1, recurrences) + 1);
+    return baseScore * decay * recurrenceBoost;
+}
+function countMistakePatternEntries(filePath, threshold) {
+    if (!(0, fs_1.existsSync)(filePath))
+        return { active: 0, dormant: 0 };
+    try {
+        const content = (0, fs_1.readFileSync)(filePath, 'utf8');
+        const lines = content.split('\n');
+        let active = 0;
+        let dormant = 0;
+        let inEntry = false;
+        let currentSeverity = null;
+        let currentLastSeen = '';
+        let currentRecurrences = 1;
+        const processCurrentEntry = () => {
+            if (!currentSeverity)
+                return;
+            const score = computeEffectiveScore(currentSeverity, currentLastSeen, currentRecurrences, 'mistake-patterns');
+            if (score >= threshold)
+                active++;
+            else
+                dormant++;
+        };
+        for (const line of lines) {
+            const headerMatch = line.match(/^## \[(P-HIGH|P-MED|P-LOW)\]/);
+            if (headerMatch) {
+                processCurrentEntry();
+                inEntry = true;
+                currentSeverity = headerMatch[1];
+                currentLastSeen = '';
+                currentRecurrences = 1;
+                continue;
+            }
+            if (!inEntry)
+                continue;
+            const lastSeenMatch = line.match(/^\*\*Last seen\*\*:\s*(.+)/);
+            if (lastSeenMatch) {
+                currentLastSeen = lastSeenMatch[1].trim();
+                continue;
+            }
+            const recurrenceMatch = line.match(/^\*\*Recurrences\*\*:\s*(\d+)/);
+            if (recurrenceMatch) {
+                currentRecurrences = parseInt(recurrenceMatch[1], 10);
+            }
+        }
+        processCurrentEntry();
+        return { active, dormant };
+    }
+    catch {
+        return { active: 0, dormant: 0 };
+    }
+}
+function readFrontmatter(content) {
+    const match = content.match(/^---\r?\n([\s\S]*?)\r?\n---/);
+    if (!match)
+        return {};
+    const frontmatter = {};
+    for (const line of match[1].split(/\r?\n/)) {
+        const separator = line.indexOf(':');
+        if (separator === -1)
+            continue;
+        const key = line.slice(0, separator).trim();
+        const value = line.slice(separator + 1).trim();
+        if (key) {
+            frontmatter[key] = value;
+        }
+    }
+    return frontmatter;
+}
+function isUnsynthesizedRetrospective(filePath) {
+    try {
+        const content = (0, fs_1.readFileSync)(filePath, 'utf8');
+        const frontmatter = readFrontmatter(content);
+        const synthesized = frontmatter.synthesized?.toLowerCase();
+        return synthesized !== 'true' && synthesized !== 'yes';
+    }
+    catch {
+        return false;
+    }
+}
+function buildLearningContextSection(workspaceRoot, userId, forJob) {
+    const learningsBase = getLearningsBase(workspaceRoot);
+    const threshold = getScoreThreshold(workspaceRoot);
+    const l2MistakePath = (0, path_1.join)(learningsBase, 'org-mistake-patterns.md');
+    const l2PrefPath = (0, path_1.join)(learningsBase, 'org-preferences.md');
+    const l2CoachPath = (0, path_1.join)(learningsBase, 'org-manager-coaching.md');
+    const l2MistakePresent = (0, fs_1.existsSync)(l2MistakePath);
+    const l2PrefPresent = (0, fs_1.existsSync)(l2PrefPath);
+    const l2CoachPresent = (0, fs_1.existsSync)(l2CoachPath);
+    const l2MistakeCounts = l2MistakePresent ? countMistakePatternEntries(l2MistakePath, threshold) : null;
+    const l1MistakePath = (0, path_1.join)(learningsBase, `${userId}-mistake-patterns.md`);
+    const l1PrefPath = (0, path_1.join)(learningsBase, `${userId}-preferences.md`);
+    const l1CoachPath = (0, path_1.join)(learningsBase, `${userId}-manager-coaching.md`);
+    const l1MistakePresent = (0, fs_1.existsSync)(l1MistakePath);
+    const l1PrefPresent = (0, fs_1.existsSync)(l1PrefPath);
+    const l1CoachPresent = (0, fs_1.existsSync)(l1CoachPath);
+    const l1MistakeCounts = l1MistakePresent ? countMistakePatternEntries(l1MistakePath, threshold) : null;
+    let l0CoachingCount = 0;
+    const rawPath = (0, path_1.join)(learningsBase, 'raw');
+    if ((0, fs_1.existsSync)(rawPath)) {
+        try {
+            l0CoachingCount = (0, fs_1.readdirSync)(rawPath).filter(f => f.startsWith(`${userId}-`)).length;
+        }
+        catch {
+            // Ignore read failures.
+        }
+    }
+    let l0RetroCount = 0;
+    const retrospectivesPath = (0, path_1.join)(workspaceRoot, 'docs', 'retrospectives');
+    if ((0, fs_1.existsSync)(retrospectivesPath)) {
+        try {
+            l0RetroCount = (0, fs_1.readdirSync)(retrospectivesPath)
+                .filter(f => f.startsWith(`${userId}-`) && f.endsWith('.md'))
+                .filter(f => isUnsynthesizedRetrospective((0, path_1.join)(retrospectivesPath, f))).length;
+        }
+        catch {
+            // Ignore read failures.
+        }
+    }
+    const hasL2 = l2MistakePresent || l2PrefPresent || l2CoachPresent;
+    const hasL1 = l1MistakePresent || l1PrefPresent || l1CoachPresent;
+    const hasContent = hasL2 || hasL1 || l0CoachingCount > 0 || l0RetroCount > 0;
+    if (!hasContent)
+        return '';
+    let section = forJob
+        ? '\n\n## Learning Context for This Job\n\n'
+        : '\n\n## Learning Context\n\n';
+    if (hasL2) {
+        section += '### L2 - Org patterns\n';
+        if (l2MistakePresent)
+            section += `\`${LEARNINGS_REL}/org-mistake-patterns.md\`  (entries above score threshold)\n`;
+        if (l2PrefPresent)
+            section += `\`${LEARNINGS_REL}/org-preferences.md\`  (all entries)\n`;
+        if (l2CoachPresent)
+            section += `\`${LEARNINGS_REL}/org-manager-coaching.md\`  (all entries)\n`;
+        if (l2MistakeCounts && l2MistakeCounts.dormant > 0) {
+            section += `Dormant: ${l2MistakeCounts.dormant} org pattern${l2MistakeCounts.dormant !== 1 ? 's' : ''} below threshold\n`;
+        }
+        section += '\n';
+    }
+    if (hasL1) {
+        section += '### L1 - Your patterns\n';
+        if (l1PrefPresent)
+            section += `\`${LEARNINGS_REL}/${userId}-preferences.md\`  (all entries)\n`;
+        if (l1CoachPresent)
+            section += `\`${LEARNINGS_REL}/${userId}-manager-coaching.md\`  (all entries)\n`;
+        if (l1MistakePresent)
+            section += `\`${LEARNINGS_REL}/${userId}-mistake-patterns.md\`  (entries above score threshold)\n`;
+        if (l1MistakeCounts && l1MistakeCounts.dormant > 0) {
+            section += `Dormant: ${l1MistakeCounts.dormant} personal pattern${l1MistakeCounts.dormant !== 1 ? 's' : ''} below threshold\n`;
+        }
+        section += '\n';
+    }
+    if (l0CoachingCount > 0 || l0RetroCount > 0) {
+        section += '### L0 - Your unprocessed signals\n';
+        if (l0CoachingCount > 0) {
+            section += `${l0CoachingCount} coaching moment${l0CoachingCount !== 1 ? 's' : ''} in \`${LEARNINGS_REL}/raw/${userId}-*\`\n`;
+        }
+        if (l0RetroCount > 0) {
+            section += `${l0RetroCount} retrospective${l0RetroCount !== 1 ? 's' : ''} in \`docs/retrospectives/${userId}-*\` with \`synthesized: false\` or missing\n`;
+        }
+        section += '\n';
+    }
+    const totalL0 = l0CoachingCount + l0RetroCount;
+    if (forJob) {
+        if (hasL2 || hasL1) {
+            section += 'Use the relevant patterns, preferences, and coaching signals in this job.\n';
+        }
+        if (totalL0 >= 5) {
+            section += '\n';
+            section += `Warning: ${totalL0} unprocessed signals pending. Consider running \`end-of-day-debrief\` before starting today's work.\n`;
+        }
+    }
+    else {
+        section += 'Use this synthesized learning context throughout the session.\n';
+        if (totalL0 >= 5) {
+            section += '\n';
+            section += `Warning: synthesis overdue with ${totalL0} unprocessed signals.\n`;
+            section += 'Run `end-of-day-debrief` before starting today\'s work.\n';
+        }
+    }
+    return section;
+}