innovation-catalyst
由 autogame-17 提供
Notes & PKM简介
Analyzes system state (memory, tools, events) and generates strategic innovation proposals to break evolution.
描述
该技能的功能
Analyzes an agent system's memory, tools, and recent events to detect evolution stagnation. When progress plateaus or tasks keep failing repeatedly, it generates actionable proposals such as new skills to acquire, optimizations to apply, or structural changes to make. Results are sent as a formatted Feishu card report.
Combines stall detection, gap analysis, and proposal generation in one automated step instead of requiring manual system audits between evolution cycles.
何时使用
- Detecting recurring task failures in an autonomous agent loop
- Identifying skill category gaps when an agent stops improving
- Generating new skill suggestions after a stable success plateau
- Sending system health summaries to a Feishu team channel
- Seeding the next evolution cycle with targeted proposals
示例工作流
User asks: 'The agent hasn't made progress in several cycles, what should it do next?'
- 1
Reads MEMORY.md, TOOLS.md, and RECENT_EVENTS.md to assess current system state
- 2
Identifies under-represented skill categories via gap analysis
- 3
Detects blocked tasks or repeated failures using stall detection
- 4
Generates actionable proposals: new skills, optimizations, structural changes
- 5
Sends a rich card report to the specified Feishu user or group
A Feishu card listing prioritized innovation proposals and detected stall points for the agent to act on in the next cycle
前置要求
- Feishu account with bot messaging permissions or webhook access
- Feishu user or group ID (ou_xxx format) to target the report
安装
clawhub install autogame-17/innovation-catalyst运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。