inspirai-evo

What This Skill Does

The inspirai-evo skill is an advanced autonomous workflow analyzer designed to help OpenClaw agents achieve self-evolution. It acts as an automated 'retro-spectator' that continuously monitors your development process for friction points, such as repetitive trial-and-error loops, frequent workflow interruptions, and high code churn. By detecting these patterns, it generates actionable analysis reports and guides the agent toward structured improvements in isolated sessions, ensuring that your primary workflow remains non-blocking and efficient.

Installation

To add this capability to your agent, use the OpenClaw installation command:

clawhub install openclaw/skills/skills/alexxxiong/inspirai-evo

Ensure that your environment allows for local file system access, as the skill maintains .evo-state.json files and creates markdown-based reports in the docs/evo-reports directory.

Use Cases

1. Refactoring Productivity: When you find yourself reverting changes multiple times or fixing the same bug repeatedly, inspirai-evo identifies the pattern and suggests permanent architectural refactoring.
2. Workflow Bottleneck Analysis: Use it to determine why certain tasks (like environment setup or testing) result in frequent interruptions or context switching.
3. Long-term Growth: Use it to track cross-project patterns, allowing your agent to learn from historical mistakes across different codebases.

Example Prompts

• "/evo --status"
• "I feel like I'm stuck in a loop with these tests, run /evo to see if we can optimize the workflow."
• "/evo --continue"

Tips & Limitations

• Thresholding: The skill follows a 3-strike rule for signal detection, meaning it will only prompt you to take action once a pattern repeats enough times to be statistically significant, preventing unnecessary interruptions.
• Isolated Sessions: Always use the --continue mode or the suggested isolated session workflow to perform heavy refactoring. This keeps the agent's main state clean and focused on task delivery.
• Manual Intervention: The skill provides recommendations but requires your final confirmation before creating or applying code-level changes, ensuring you maintain full control over your project's evolution.