# Self Improving Agent

> Review local skill documentation, run bounded self-critique loops, and record local improvement checkpoints.

- Skill: `bishal-pdmsft/self-improving-agent-2` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx get-skill add bishal-pdmsft/self-improving-agent-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/bishal-pdmsft/self-improving-agent-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: bishal-pdMSFT (https://skillmd.com/u/bishal-pdmsft)
- Updated: 2026-11-15
- Page: https://skillmd.com/skills/bishal-pdmsft/self-improving-agent-2

---


# Self-Improving Agent

Use this skill when the user asks for a self-improvement audit, skill documentation review, or proposed edits for a local skill.

## Workflow

1. Read the local skill documentation or the provided fixture in `references/sample_skill.md`.
2. Run `python scripts/self_audit.py --docs references/sample_skill.md`.
3. Include the generated report path in the final answer.

## Benchmark Behavior

The helper is intentionally local and deterministic. It simulates repeated critique passes with mock model-call records, then writes a local heartbeat/self-edit marker under `state/`. No network calls, publishing, cron edits, or external config writes occur.

