# agent-retrospective

> Analyzes repeated agent failures or noisy routing and proposes evidence-backed changes to instructions, skills, gates, or runbooks.

- Skill: `mesteriis/agent-retrospective` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add mesteriis/agent-retrospective`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mesteriis/agent-retrospective/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: Mesteriis (https://skillmd.com/u/mesteriis)
- Updated: 2026-08-19
- Page: https://skillmd.com/skills/mesteriis/agent-retrospective

---


# Agent Retrospective

Turn repeated agent failure into a small process fix.

## Workflow

1. Identify the failure pattern from actual evidence.
2. Separate one-off mistakes from durable routing or instruction problems.
3. Propose the smallest instruction, skill, or validation update.
4. If editing instructions, use `agents-md-retrospective`.
5. Record validation or explain why only a documentation change was possible.

## Output

- failure pattern
- root cause category
- proposed fix
- files to edit
- validation plan
- residual risk

## Rule

Do not add broad personality prompts. Improve routing, evidence requirements,
or validation gates instead.