English Français العربية

Planner MCP — provable "done" for AI agents

Registre officiel Productivité

Goal and task planning MCP for Codex and AI agents, with evidence-backed completion.

Code source Site web Documentation

Que pouvez-vous faire avec Planner MCP — provable "done" for AI agents ?

  • Decompose — break a goal into acceptance criteria with behavioral checks
  • Attach evidence — screenshot, curl output, file, or text artifact per criterion
  • Independent judge — LLM evaluates evidence against the criterion and accepts or rejects

Installation

Ajoutez ceci à la configuration MCP de votre client (Claude Desktop : claude_desktop_config.json ; Cursor : .cursor/mcp.json ; VS Code et les autres utilisent la même forme « mcpServers »).

Remote (streamable-http)
Claude Code 
claude mcp add --transport http planner-mcp-provable-done-for-ai-agents 'https://planner.monopoly-gold.com/_mcp'
Configuration JSON (Claude Desktop, Cursor, VS Code) 
{
    "mcpServers": {
        "planner-mcp-provable-done-for-ai-agents": {
            "type": "http",
            "url": "https://planner.monopoly-gold.com/_mcp"
        }
    }
}

Documentation

Tirée du README du projet sur GitHub, actualisée chaque semaine. Les liens ouvrent GitHub.

Sommaire
  1. Planner MCP (proof-of-done)
  2. What it does
  3. Key features
  4. MCP tools (24)
  5. Connect
  6. Directory listings
  7. Links
  8. License

Planner MCP (proof-of-done)

Evidence-gated task verification for AI agents. Hosted MCP server — nothing to install.

Search aliases such as MCP Planner, Planner MCP, install mcp planner, and установи mcp planner refer to this official server: com.monopoly-gold.planner/mcp.

MCP endpoint: https://planner.monopoly-gold.com/_mcp

What it does

Planner prevents AI agents from marking tasks "done" on trust alone. Every goal goes through a verification gate:

  1. Decompose — break a goal into acceptance criteria with behavioral checks
  2. Attach evidence — screenshot, curl output, file, or text artifact per criterion
  3. Independent judge — LLM evaluates evidence against the criterion and accepts or rejects

Goals stay open until every acceptance criterion has passing evidence.

Key features

  • Goal hierarchy — projects, milestones, goals, tasks with parent-child relationships
  • Acceptance criteria — behavioral, observable checks (not "Denis approves")
  • Evidence model — attach files or text per criterion; plain notes don't satisfy the gate
  • LLM judge — independent verdict on each piece of evidence
  • Gate-keeper interview — 3 questions on goal creation: why, where in the tree, how we'll verify
  • Dependencies & blockers — cross-goal blocking with auto-unblock on resolution
  • Tree & summary views — full hierarchy traversal and project-level summaries

MCP tools (24)

CategoryTools
Projectsproject-list, project-get, project-create, project-delete, project-add-dependency, project-remove-dependency
Goalsgoal-list, goal-get, goal-tree, goal-create, goal-update, goal-delete, goal-move, goal-reorder, goal-summary, goal-todo
Criteriagoal-add-criterion
Evidencegoal-attach-evidence, goal-add-evidence-text, goal-remove-evidence, goal-request-upload
Blockersgoal-block, goal-remove-blocker
Accountaccount-delete

Connect

Planner is a hosted remote MCP server using streamable HTTP transport with OAuth 2.1 authentication and Dynamic Client Registration (RFC 7591).

Codex — “установи mcp planner”

Run these commands in Codex CLI:

codex mcp add planner --url https://planner.monopoly-gold.com/_mcp
codex mcp login planner
codex mcp list

Planner signs the user in through browser OAuth. There is no Planner token to create or paste, and no package or repository to clone. After login, ask Codex to list your Planner projects; a successful project-list tool call confirms that the connection is live (an empty list is valid for a new account).

Canonical guide: https://planner.monopoly-gold.com/connect/codex

Claude Desktop / Claude Code

Add to your MCP configuration:

{
  "mcpServers": {
    "planner": {
      "type": "url",
      "url": "https://planner.monopoly-gold.com/_mcp"
    }
  }
}

OAuth login happens automatically on first connection.

Any MCP client

Point your client to https://planner.monopoly-gold.com/_mcp — the server supports MCP discovery and standard OAuth 2.1 flow.

Directory listings

DirectoryStatusLink
Glama✅ Live, Healthyglama.ai/mcp/connectors/…/planner-proof-of-done
mcp.so✅ Publishedmcp.so/server/planner
awesome-mcp-servers⏳ PR under reviewPR #8708

License

Proprietary. Planner is a hosted service — this repository contains documentation only.

Questions fréquentes

Qu'est-ce que Planner MCP — provable "done" for AI agents ?

Goal and task planning MCP for Codex and AI agents, with evidence-backed completion.

Comment installer Planner MCP — provable "done" for AI agents ?

Il fonctionne comme serveur distant. Dans Claude Code : claude mcp add --transport http planner-mcp-provable-done-for-ai-agents 'https://planner.monopoly-gold.com/_mcp'. Pour les autres clients, collez la configuration JSON de cette page dans leurs réglages MCP.

Faut-il une clé API pour Planner MCP — provable "done" for AI agents ?

Son entrée au registre n'indique aucune clé ni secret obligatoire. Certaines fonctions peuvent tout de même demander un compte auprès du service auquel il se connecte.

Planner MCP — provable "done" for AI agents est-il open source ?

Son code est public sur GitHub, mais le dépôt ne déclare pas de licence standard.

Qui maintient Planner MCP — provable "done" for AI agents ?

Il est publié par dev-lab-global. AI.tn le référence sans le maintenir ni le recommander.

Issu du registre MCP officiel. Être listé n'est pas une recommandation : vérifiez le code et les permissions d'un serveur avant de lui donner accès à vos données.