Lessons learned
Plugin with opinionated set of Claude Code agents nad skills
npx -y skills add lklimek/claudius --skill lessons-learnedAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use when extracting learnings, saving lessons, or capturing reusable knowledge from the session. Also appropriate before presenting a plan, after notable events (bugs, wrong approaches corrected), and as final task when work is complete.
SKILL.md
1.7 KB, as published. Nobody here has run it
Lessons Learned
Extract, qualify, and persist reusable knowledge from the current session.
Source of Truth
!cat ${CLAUDE_SKILL_DIR}/../../references/source-of-truth.md
Phase 1 — Gather
Scan the conversation for items worth remembering. Use the categories, quality gate, and examples from the Source of Truth above.
Collect as a numbered list. Search existing knowledge (memcan:recall) and drop duplicates.
Tone: factual, third-person, present tense. Pattern: "[Subject]: [what/what to do] — [why/context]". Apply the Source of Truth Authoring rules — self-contained phrasing (no pronouns/anaphora; inline the subject) and timeless present tense (strip "now"/"previously"/"currently"/"no longer").
Opportunistic Cleanup
During dedup searches, if existing memories fail the quality gate, update or delete them.
Phase 2 — Save
For each qualified item:
- Assign scope: global (cross-project, omit
project) or project-scoped (setprojectto git remote origin repo name) - Assign type: lesson, decision, or preference
- Invoke
memcan:rememberskill to persist each item - If memcan is unavailable, report findings but note they could not be persisted
Log each save: scope, type, one-line summary. Report total count.
IMPORTANT: This skill does NOT call memcan MCP tools directly. It invokes memcan:remember for saves and memcan:recall for searches. Claudius owns the classification logic; memcan is the execution layer.