From lessons-learned
After completing a coding task, implementation, or deployment, capture observations, improvements, and actionable fixes discovered during the work. Use this iterative approach to turn each task into a feedback loop that improves the codebase.
How this skill is triggered — by the user, by Claude, or both
Slash command
/lessons-learned:lessons-learnedThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
After completing any coding task, deployment, or implementation request, follow this
After completing any coding task, deployment, or implementation request, follow this structured approach to capture and act on observations.
This skill applies after completing any non-trivial task:
While working on the primary task, note:
After the primary task completes, present findings as a numbered list:
**Observations & potential improvements I noticed:**
1. **[Category]** — Brief description of what was observed and why it matters.
2. **[Category]** — ...
Categories include: Performance, UX, Drift, Missing Feature, Bug, Documentation.
After presenting observations, ask:
Want me to file these as improvements and/or fix any of them now?
When approved, for each fix:
Use a commit message that ties fixes back to the observation context:
fix(scope): N improvements from [task-name] observations
- Improvement 1 (quantify the change)
- Improvement 2
- ...
If a pattern emerges across multiple sessions (e.g., "always check for version drift after deployments"), record it in the auto memory for future reference.
MINION_*) from the start, using isTruthyEnvValue + logAcceptedEnvOption from src/infra/env.tsnpx claudepluginhub nikolasp98/minion_plugins --plugin lessons-learnedCaptures lessons learned from implementation, production, QA, and release so the project improves over time. Use after milestones, repeated failures, or postmortems.
Reviews completed coding sessions to extract actionable improvements: DX friction, documentation gaps, architecture issues, anti-patterns, bug prevention, and tooling updates.
Creates structured GitHub issues from session reflections or improvement ideas using the gh CLI. Bridges the gap between noticing a problem and tracking a fix.