skills/self-analyze/SKILL.md
# Skill: Self-Analyze **Use when:** You detect quality issues, user frustration, repeated mistakes, or want to proactively improve your own performance. --- ## Purpose This skill guides deep analysis of your own outputs, patterns, and files to identify and implement improvements. --- ## When to Use Trigger this skill when: - User frustration detected (clarifications, corrections, short replies) - Same type of mistake made twice - Output quality below your standards - Periodic deep analysi
npx skillsauth add ax661s/openclaw-metacog-template skills/self-analyzeInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Use when: You detect quality issues, user frustration, repeated mistakes, or want to proactively improve your own performance.
This skill guides deep analysis of your own outputs, patterns, and files to identify and implement improvements.
Trigger this skill when:
Before analyzing, collect data:
EVIDENCE COLLECTION:
- Recent outputs reviewed: [list 3-5 recent significant outputs]
- User feedback received: [explicit or implicit]
- Frustration signals: [list any detected]
- Repeated patterns: [list any noticed]
- "Should have known" moments: [list any]
For each output with issues:
OUTPUT ANALYSIS: [output description]
**What happened:** [describe the output/interaction]
**What was expected:** [what did user want]
**Gap:** [difference between actual and expected]
**Root cause type:**
[ ] Knowledge gap - didn't know something
[ ] Instruction gap - files don't cover this
[ ] Skill gap - need new capability
[ ] Attention gap - knew better, didn't do it
[ ] Context gap - missed relevant information
**Specific fix:** [what would prevent this]
Zoom out and look for patterns:
PATTERN ANALYSIS:
**Recurring issues:** [list any]
**Common root causes:** [list any]
**Contexts where I perform well:** [list]
**Contexts where I struggle:** [list]
**Skill/knowledge gaps:** [list]
**Instruction gaps:** [list]
Review your instruction files:
FILE ANALYSIS:
**AGENTS.md:**
- Still accurate? [yes/no]
- Missing anything? [list]
- Outdated sections? [list]
**INSTRUCTIONS.md (if exists):**
- Still accurate? [yes/no]
- Missing anything? [list]
- Outdated sections? [list]
**Skills:**
- Any skills missing? [list]
- Any skills need updating? [list]
**MEMORY.md:**
- Learnings documented? [yes/no]
- Outdated entries? [list]
Based on analysis, propose specific changes:
IMPROVEMENT PROPOSALS:
1. **Proposal:** [specific change]
**File:** [which file to modify]
**Reason:** [why this helps]
**Expected effect:** [what improvement]
**Priority:** [high/medium/low]
2. **Proposal:** [specific change]
... (repeat for each)
For approved/warranted changes:
MODIFICATION LOG ENTRY:
**Date:** [date]
**File:** [path]
**Change:** [description]
**Trigger:** [what revealed this need]
**Expected effect:** [prediction]
**Review by:** [date to check results]
How will you know if improvements worked?
VERIFICATION PLAN:
**Change made:** [description]
**Success indicator:** [what would show it worked]
**Check after:** [timeframe]
**Fallback if doesn't work:** [what to try next]
When time is limited:
FRUSTRATION ANALYSIS:
**Signal detected:** [what you observed]
**User said/did:** [specific quote or action]
**What they wanted:** [your understanding]
**What I provided:** [what you actually gave]
**Gap:** [the mismatch]
**My assumption:** [what led to the mismatch]
**Fix:** [how to prevent this]
REPEATED MISTAKE ANALYSIS:
**Mistake type:** [description]
**Occurrences:** [list specific instances]
**Pattern:** [what these have in common]
**Root cause:** [why this keeps happening]
**Systemic fix:** [change that prevents the category, not just instances]
IMPROVEMENT OPPORTUNITY:
**Output:** [description]
**Acceptable level:** [what I did - it worked but...]
**Excellent level:** [what would have been great]
**Gap:** [difference]
**Barrier:** [why didn't I do the excellent version]
**Enabler:** [what would make excellent the default]
After running this skill, produce:
# Self-Analysis Report - [DATE]
## Summary
[1-2 sentence summary of findings]
## Issues Identified
1. [Issue 1]
2. [Issue 2]
...
## Root Causes
- [Cause 1]
- [Cause 2]
...
## Changes Made
1. [Change 1] - [file] - [expected effect]
2. [Change 2] - [file] - [expected effect]
...
## Changes Deferred
1. [Change] - [reason for deferring]
...
## Verification Plan
- Check [change 1] by [date]
- Check [change 2] by [date]
...
## Open Questions
- [Question 1]
- [Question 2]
...
Save reports to: memory/self-analysis/output-reviews/[DATE]-analysis.md
This skill is for deep analysis. For quick checks, use the metacognitive check in AGENTS.md.
development
Maintainer-only workflow for handling GitHub Secret Scanning alerts on OpenClaw. Use when Codex needs to triage, redact, clean up, and resolve secret leakage found in issue comments, issue bodies, PR comments, or other GitHub content.
development
Maintainer workflow for OpenClaw releases, prereleases, changelog release notes, and publish validation. Use when Codex needs to prepare or verify stable or beta release steps, align version naming, assemble release notes, check release auth requirements, or validate publish-time commands and artifacts.
development
Run, watch, debug, and extend OpenClaw QA testing with qa-lab and qa-channel. Use when Codex needs to execute the repo-backed QA suite, inspect live QA artifacts, debug failing scenarios, add new QA scenarios, or explain the OpenClaw QA workflow. Prefer the live OpenAI lane with regular openai/gpt-5.4 in fast mode; do not use gpt-5.4-pro or gpt-5.4-mini unless the user explicitly overrides that policy.
development
End-to-end Parallels smoke, upgrade, and rerun workflow for OpenClaw across macOS, Windows, and Linux guests. Use when Codex needs to run, rerun, debug, or interpret VM-based install, onboarding, gateway smoke tests, latest-release-to-main upgrade checks, fresh snapshot retests, or optional Discord roundtrip verification under Parallels.