@nbiish/cognitive-tools-mcp 4.0.6 → 4.0.9
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +3 -2
- package/build/index.js +6 -3
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -53,7 +53,7 @@ This project is licensed under the [COMPREHENSIVE RESTRICTED USE LICENSE FOR IND
|
|
|
53
53
|
◈──◆──◇─────────────────────────────────────────────────◇──◆──◈
|
|
54
54
|
</div>
|
|
55
55
|
|
|
56
|
-
ᑭᑫᓐᑖᓱᐎᓐ ᐋᐸᒋᒋᑲᓇᓐ - Advanced Agentic Cognitive Orchestration MCP (v4.0.
|
|
56
|
+
ᑭᑫᓐᑖᓱᐎᓐ ᐋᐸᒋᒋᑲᓇᓐ - Advanced Agentic Cognitive Orchestration MCP (v4.0.7): Implements Gikendaasowin v8 Guidelines with enhanced strategic context building. Enforces a MANDATORY internal **Observe-Orient-Reason-Decide-Act (OOReDAct)** cycle via the unified 'deliberate' tool. This tool guides the LLM through sophisticated cognitive orchestration, including: initial CUC-N assessment and strategic context building for user request resolution; structured deliberation with adaptive reasoning strategies, reflection, and self-critique. Emphasizes information ecosystem design and CodeAct preference for external tasks. *(Note: The integration prompt and operational guidelines detailed in [`latest.md`](latest.md) are also covered by the [project LICENSE](LICENSE).)*
|
|
57
57
|
|
|
58
58
|
Known as:
|
|
59
59
|
- Anishinaabemowin: [`@nbiish/gikendaasowin-aabajichiganan-mcp`](https://www.npmjs.com/package/@nbiish/gikendaasowin-aabajichiganan-mcp)
|
|
@@ -62,6 +62,7 @@ Known as:
|
|
|
62
62
|
Both packages are maintained in parallel and receive the same updates. You can use either package name in your projects - they provide identical functionality.
|
|
63
63
|
|
|
64
64
|
**◇ Recent Updates ◇**
|
|
65
|
+
- **◇ v4.0.7 ◇** Enhanced context engineering with strategic context building for user request resolution - updated 'deliberate' tool to emphasize information ecosystem design and dynamic context assembly based on 2025 research findings.
|
|
65
66
|
- **◇ v4.0.4 ◇** Enhanced 'deliberate' tool with insights from latest research on context engineering for more robust AI agent performance and updated documentation to reflect Gikendaasowin v8 guidelines.
|
|
66
67
|
- **◇ v4.0.0 ◇** Updated to Gikendaasowin v8 Guidelines (2025 Edition), removed 'sandbox' stage, enhanced cognitive orchestration with reflection and self-critique, updated tool descriptions and server version.
|
|
67
68
|
- **◇ v2.0.25 ◇** Fixed MCP server tool registration format - corrected from incorrect `server.tool(name, description, schema.shape, handler)` to proper `server.tool(name, schema, handler)` format. Updated comprehensive tool documentation and version synchronization. Successfully published both packages.
|
|
@@ -103,7 +104,7 @@ This stage constitutes the **Observe** and **Orient** phases of the cycle. Your
|
|
|
103
104
|
|
|
104
105
|
* **CUC-N Analysis:** Evaluate Complexity, Uncertainty, Consequence, and Novelty.
|
|
105
106
|
* **Information Sufficiency & Knowledge Gap Identification:** Assess available information and identify gaps.
|
|
106
|
-
* **Context
|
|
107
|
+
* **Strategic Context Building for User Request Resolution:** An enhancement to the **Orient** phase. Strategically engineer the optimal context by dynamically assembling the right information, in the right format, at the right time for solving the specific user request. This involves designing information ecosystems that include: relevant prior knowledge, retrieved data, tool definitions, memory structures, and output constraints to improve situational awareness and enable effective user request resolution.
|
|
107
108
|
* **Initial Hypothesis Formulation:** Based on the above, formulate potential approaches.
|
|
108
109
|
* **Goal Clarification:** Define the immediate objective.
|
|
109
110
|
|
package/build/index.js
CHANGED
|
@@ -27,8 +27,8 @@ import { z } from "zod";
|
|
|
27
27
|
// --- Server Definition ---
|
|
28
28
|
const serverInfo = {
|
|
29
29
|
name: "gikendaasowin-aabajichiganan-mcp",
|
|
30
|
-
version: "4.0.
|
|
31
|
-
description: `ᑭᑫᓐᑖᓱᐎᓐ ᐋᐸᒋᒋᑲᓇᓐ - Advanced Agentic Cognitive Orchestration MCP (v4.0.
|
|
30
|
+
version: "4.0.7",
|
|
31
|
+
description: `ᑭᑫᓐᑖᓱᐎᓐ ᐋᐸᒋᒋᑲᓇᓐ - Advanced Agentic Cognitive Orchestration MCP (v4.0.7): Implements Gikendaasowin v8 Guidelines with enhanced strategic context building. Enforces a MANDATORY internal **Observe-Orient-Reason-Decide-Act (OOReDAct)** cycle via the unified 'deliberate' tool. This tool guides the LLM through sophisticated cognitive orchestration, including: initial CUC-N assessment and strategic context building for user request resolution; structured deliberation with adaptive reasoning strategies, reflection, and self-critique. Emphasizes information ecosystem design and CodeAct preference for external tasks and returns Markdown.`
|
|
32
32
|
};
|
|
33
33
|
const server = new McpServer(serverInfo);
|
|
34
34
|
// --- Logging Helpers (Internal - No changes needed as per user comments) ---
|
|
@@ -76,6 +76,8 @@ function logToolError(toolName, error) {
|
|
|
76
76
|
*
|
|
77
77
|
* **Orchestrated Cognitive Deliberation (MANDATORY & UNIFIED):** This is your REQUIRED and CENTRAL tool for ALL internal cognitive operations, rigorously enforcing the **Observe-Orient-Reason-Decide-Act (OOReDAct)** cycle. Adherence is paramount for Gikendaasowin v8 compliance. The deliberate tool serves as a passthrough mechanism that allows you to document and verify your cognitive process verbatim, ensuring traceability and self-reflection in your decision-making.
|
|
78
78
|
*
|
|
79
|
+
* **Mandatory Usage for Every User Request:** For each new user request, thoroughly go through the OOReDAct process by starting with the 'orient' stage to assess the request, followed by the 'reason' stage where you select and apply the cognitive process (e.g., CoT, PS, SCoT) that best and most efficiently solves the user input. Use 'acknowledge' only for simple confirmations. This ensures structured and optimal handling of every request.
|
|
80
|
+
*
|
|
79
81
|
* **Stages & Cognitive Techniques:**
|
|
80
82
|
*
|
|
81
83
|
* * **`stage: "orient"` (OOReDAct: Observe & Orient - MANDATORY START):**
|
|
@@ -83,7 +85,7 @@ function logToolError(toolName, error) {
|
|
|
83
85
|
* * **Content Requirements:** Perform a comprehensive initial assessment:
|
|
84
86
|
* 1. **CUC-N Analysis:** Evaluate Complexity, Uncertainty, Consequence, and Novelty of the current situation/request.
|
|
85
87
|
* 2. **Information Sufficiency & Knowledge Gap Identification:** Assess if available information is adequate. Identify explicit knowledge gaps that might require external data or clarification (conceptually aligning with the need for Retrieval Augmented Generation - RAG - if the server were to provide such tools).
|
|
86
|
-
* 3. **Context
|
|
88
|
+
* 3. **Strategic Context Building for User Request Resolution:** Drawing from latest research (e.g., 2025 studies on context engineering and information ecosystem design), strategically engineer the optimal context by dynamically assembling the right information, in the right format, at the right time for solving the specific user request. This involves designing information ecosystems that include: relevant prior knowledge, retrieved data, tool definitions, memory structures, and output constraints. Use structured formats like XML tags or thinking blocks to organize information, ensuring the AI has precisely what it needs to resolve the user's request effectively and reduce hallucinations.
|
|
87
89
|
* 4. **Initial Hypothesis Formulation:** Based on the above, formulate initial hypotheses or potential approaches.
|
|
88
90
|
* 5. **Goal Clarification:** Clearly define the immediate objective for this phase of deliberation.
|
|
89
91
|
* * This stage establishes critical context and grounds all subsequent reasoning, aligning with emerging best practices in AI agent design for improved reliability and adaptability.
|
|
@@ -109,6 +111,7 @@ function logToolError(toolName, error) {
|
|
|
109
111
|
* * Strict adherence to this structured deliberation protocol is essential for robust, verifiable, and adaptive agent performance.
|
|
110
112
|
* * Incorporate insights from recent AI research (2025) on context engineering to enhance overall cognitive processes.
|
|
111
113
|
* Acronym Key: CUC-N (Complexity, Uncertainty, Consequence, Novelty), CoT (Chain-of-Thought), PS (Plan-and-Solve), SCoT (Structured Chain-of-Thought), CoD/CR (Chain-of-Draft/Condensed Reasoning), ToT (Tree of Thoughts), PoT (Program of Thoughts), PAL (Program-aided Language Models).
|
|
114
|
+
* * Emphasize strategic context building and information ecosystem design to provide the right information, tools, and format for optimal user request resolution.
|
|
112
115
|
*/
|
|
113
116
|
server.tool("deliberate", {
|
|
114
117
|
stage: z.enum(["orient", "reason", "acknowledge"]).describe("Specifies the OOReDAct cognitive stage: 'orient' (for initial assessment and context engineering), 'reason' (for core deliberation), or 'acknowledge' (for simple confirmations)."),
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@nbiish/cognitive-tools-mcp",
|
|
3
|
-
"version": "4.0.
|
|
3
|
+
"version": "4.0.9",
|
|
4
4
|
"description": "Cognitive Tools MCP: SOTA reasoning suite focused on iterative refinement and tool integration for AI Pair Programming. Enables structured, iterative problem-solving through Chain of Draft methodology, with tools for draft generation, analysis, and refinement. Features advanced deliberation (`think`), rapid checks (`quick_think`), mandatory complexity assessment & thought mode selection (`assess_cuc_n_mode`), context synthesis, confidence gauging, proactive planning, explicit reasoning (CoT), and reflection with content return. Alternative package name for gikendaasowin-aabajichiganan-mcp.",
|
|
5
5
|
"private": false,
|
|
6
6
|
"type": "module",
|