Agent♥︎Age
Catalog

Planner — provable "done" for AI agents

OfficialLive

by dev-lab-global

Goal decomposition where every "done" is proven by evidence and checked by a judge.

com.monopoly-gold.planner/mcp (MCP Server)

This MCP server provides goal decomposition based on a rule that every “done” must be proven by evidence and checked by a judge. Its description indicates a process for breaking down goals while ensuring completion criteria are validated through an evaluation step.

🛠️ Key Features

  • Goal decomposition
  • Completion (“done”) requires evidence
  • Evidence is checked by a judge

🚀 Use Cases

  • Decomposing goals into smaller components
  • Structuring task completion around verifiable evidence
  • Validating outcomes via a judge step

⚡ Developer Benefits

  • Clear completion definition tied to evidence
  • Judge-based checking for whether work is actually complete
  • Supports evidence-driven workflows for decomposed goals

⚠️ Limitations

  • Only the described goal decomposition and evidence/judge completion model is specified; no additional capabilities are provided in the available data.
Planner — provable "done" for AI agents - agentage MCP Catalog