learning-from-mistakes · syahiidkamil — Claude Code skill: logs bugs
learning-from-mistakes — a skill that writes a bug or mistake into a separate file after the problem is resolved. Captures the cause, fix, and lesson to avoid repeating the debugging work.
no license ★ 406 updated 2026-06-25
npx skills add syahiidkamil/Software-Engineer-AI-Agent-Atlas --skill learning-from-mistakes -a claude-codeWhat learning-from-mistakes does
The skill generates a file docs/learning-from-mistakes/<problem-name>.md after a hard bug or long debugging session. It includes four sections: symptom (what looked wrong), root cause (why it actually happened), resolution (what fixed it), and lesson (how to avoid it next time). The agent writes an entry on its own when a bug resisted solving and required guidance from the developer, or when a debugging hunt took a long time and ended in a non-obvious realization.
Before declaring a bug unsolvable, the agent scans docs/learning-from-mistakes/ — a past lesson may already name the cause. Routine fixes are skipped: the criterion is surprise. An entry is worth writing only if the cause would genuinely surprise someone, even in hindsight.
The skill works with the ATLAS repository (Software Engineer AI Agent), where the Claude Code agent manages the project. Entries accumulate locally, each as a separate file.
What it helps with
- capture a non-obvious bug cause after debugging
- preserve a lesson to avoid repeating the same mistake
- check past bugs before giving up on a problem
Example prompts for the agent
- Write a lesson from this bug into
learning-from-mistakes - Scan
docs/learning-from-mistakes/— we might have hit this before - Create an entry for this issue: symptom was timezone shifting the date by one day
How to install learning-from-mistakes
1Skills CLI (Vercel)
npx skills add syahiidkamil/Software-Engineer-AI-Agent-Atlas --skill learning-from-mistakes -a claude-code2Manually: put the files in place
~/.claude/skills/learning-from-mistakes/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/learning-from-mistakes/
1Skills CLI (Vercel)
npx skills add syahiidkamil/Software-Engineer-AI-Agent-Atlas --skill learning-from-mistakes -a codex2Manually: put the files in place
.agents/skills/learning-from-mistakes/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add syahiidkamil/Software-Engineer-AI-Agent-Atlas --skill learning-from-mistakes -a cursor2Manually: put the files in place
.cursor/skills/learning-from-mistakes/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install syahiidkamil/Software-Engineer-AI-Agent-Atlas .claude/skills/learning-from-mistakes2Manually: put the files in place
.github/skills/learning-from-mistakes/SKILL.md1Skills CLI (Vercel)
npx skills add syahiidkamil/Software-Engineer-AI-Agent-Atlas --skill learning-from-mistakes -a gemini-cli2Platform CLI
gemini skills install https://github.com/syahiidkamil/Software-Engineer-AI-Agent-Atlas --path .claude/skills/learning-from-mistakesManually: put the files in place
.devin/skills/learning-from-mistakes/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about learning-from-mistakes
How to install learning-from-mistakes in Claude Code?
Run npx skills add syahiidkamil/Software-Engineer-AI-Agent-Atlas --skill learning-from-mistakes -a claude-code, and the skill will land in ~/.claude/skills/learning-from-mistakes/SKILL.md.
What does the learning-from-mistakes skill do?
It writes a bug or mistake into a file after resolution: symptom, cause, fix, and lesson. The agent creates an entry on its own when a bug proved hard or surprising.
When does the agent create an entry?
When a bug resisted solving and required developer guidance, or when debugging took a long time and ended in a non-obvious insight. Routine fixes are not recorded.
Similar skills
All skills for Agent setup: memory, context, orchestration →learning-from-mistakes by other authors comes first.