1. Conduid
  2. Developer Tools
  3. Coaia Memory
MCP server · Developer Tools

Coaia Memory

Creative-Oriented AI Assistant Memory - MCP server with structural tension charts and advancing pattern support, based on principles by Robert Fritz

37Low

Scored 6 months ago · breakdown

About Coaia Memory

Coaia Memory is an MCP server in the Developer Tools category: creative-Oriented AI Assistant Memory - MCP server with structural tension charts and advancing pattern support, based on principles by Robert Fritz.

Install

Claude Code
claude mcp add coaia-memory -- npx -y coaia-memory
npx
npx -y coaia-memory

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Coaia Memory

Powered by Claude · Grounded in docs

I know everything about Coaia Memory. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

  • ✓README presentThe repository ships documentation.

5 checks have not been run yet.

README

COAIA Memory

MCP server for structural tension charts based on Robert Fritz's creative methodology

Extends @modelcontextprotocol/server-memory with structural tension charts - a different orientation from problem-solving, focused on bringing desired outcomes into being through structural dynamics.

Quick Start

npx coaia-memory --memory-path ./charts.jsonl

Configuration

Minimal Setup (STC Tools Only - Default)

{
  "mcpServers": {
    "coaia-memory": {
      "command": "npx",
      "args": ["-y", "coaia-memory", "--memory-path", "/path/to/charts.jsonl"]
    }
  }
}

Exposes 12 tools: list_active_charts, create_structural_tension_chart, add_action_step, remove_action_step, telescope_action_step, mark_action_complete, get_chart_progress, update_action_progress, update_current_reality, update_desired_outcome, update_action_step_title, init_llm_guidance

Full Setup (STC + Knowledge Graph Tools)

{
  "mcpServers": {
    "coaia-memory": {
      "command": "npx",
      "args": ["-y", "coaia-memory", "--memory-path", "/path/to/charts.jsonl"],
      "env": {
        "COAIA_TOOLS": "STC_TOOLS,KG_TOOLS,init_llm_guidance"
      }
    }
  }
}

Adds 9 KG tools: create_entities, create_relations, add_observations, delete_entities, delete_observations, delete_relations, search_nodes, open_nodes, read_graph

Environment Variables

Variable Description Default
COAIA_TOOLS Tool groups/names to enable (comma/space separated) STC_TOOLS,init_llm_guidance
COAIA_DISABLED_TOOLS Tools to exclude from enabled set -

Tool Groups:

  • STC_TOOLS - Structural tension chart tools (11)
  • KG_TOOLS - Knowledge graph tools (9)
  • CORE_TOOLS - Essential tools only (4): list_active_charts, create_structural_tension_chart, add_action_step, mark_action_complete

Core Concepts

Structural Tension

┌─────────────────┐                    ┌─────────────────┐
│ Current Reality │ ══ TENSION ══════> │ Desired Outcome │
│ (where you are) │                    │ (what to create)│
└─────────────────┘                    └─────────────────┘
         │                                      ▲
         │       Strategic Secondary Choices    │
         └──────────────────────────────────────┘
        (action steps understood in context of tension)
  • Desired Outcome: What you want to CREATE (not fix/solve). Specific, quantified where possible, avoiding comparative terms (more, better, less).
  • Current Reality: Honest factual assessment - objective facts only, no readiness assumptions ("ready to begin" destroys tension).
  • Structural Tension: The dynamic force between current reality and desired outcome that naturally seeks resolution. NOT a gap to fill, but a generative force.

Action Steps: NOT a To-Do List

Critical distinction: Action steps are strategic secondary choices that support the primary goal, NOT tasks to check off.

They are:

  • Understood in the context of structural tension
  • Related to each other as part of an overview
  • Strategic actions designed to enable creating the goal
  • Answers to: "If we took these steps, would we achieve this result?"

Three types of actions (use appropriately):

  1. Overview actions - Strategic steps early in process (what we put on charts)
  2. Experimental actions - Learning, exploring, "sketches before the painting"
  3. Refinement actions - Polishing near completion (too early = stifled energy)

Test question: "If we took these steps, would we achieve this result?" If No → add more steps. If Yes → done.

Goal Refinement Principles (Robert Fritz)

Principle Wrong Right
Quantify "Increased business" "5 new business clients"
No comparatives "Better health" "Very good health"
Create, not solve "Overcome weight problem" "I weigh 150 pounds"
Result, not process "Run 4 miles daily" "Well-toned, healthy body"
Specific, not vague "Improve my skills" "Mastery of Photoshop"

Creator Moment of Truth (Review Process)

When assessing progress, use this four-step process:

  1. Acknowledge: What was expected vs. what was delivered? (facts only)
  2. Analyze: How did it happen? (step-by-step, not blame)
  3. Plan: How will I do it differently next time?
  4. Feedback: How will I track the changes?

This transforms discrepancies into learning opportunities, not failures.

Usage Examples

Create a chart:

{
  "desiredOutcome": "Launch personal website",
  "currentReality": "Have domain, no design or content yet",
  "dueDate": "2025-03-01T00:00:00Z",
  "actionSteps": ["Design homepage", "Write about page", "Deploy to hosting"]
}

Add action step to existing chart:

{
  "parentChartId": "chart_1234567890",
  "actionStepTitle": "Set up CI/CD pipeline",
  "currentReality": "Manual deployment only, no automation experience"
}

Mark action complete:

{ "actionStepName": "chart_1234567890_desired_outcome" }

Development

git clone https://github.com/jgwill/coaia-memory
cd coaia-memory
npm install
npm run build

Credits

README copy captured when this server was indexed; it may lag the repository. The original is authoritative.

Questions

About Coaia Memory

How do I install Coaia Memory?

Run claude mcp add coaia-memory -- npx -y coaia-memory, then add the server to your MCP client's configuration.

Is Coaia Memory safe to use with an AI agent?

Its trust score is 37 out of 100 (low). It passes 1 of the 1 static security checks run so far. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Coaia Memory still maintained?

Conduid hasn't recorded a commit date for this repository yet. Check the repository directly for recent activity.