elliotjlt/claude-skill-potions/retrospective
Claude Code and agent skill by elliotjlt/claude-skill-potions
- community
Checking that the SKILL.md file is still in the repository.
After completing a significant task or experiment, documents what worked, what failed, and key learnings. Activates when a multi-step task finishes or when user says "done", "finished", "that worked", or asks for a summary. Failed attempts get documented first - they're read more than successes.
Install
npx vskill@latest install elliotjlt/claude-skill-potions/retrospectiveSKILL.md
Retrospective
<purpose> Knowledge capture at task completion. The Sionic AI insight: "Failed Attempts" tables get read more than any other documentation section. Success paths are nice. Failure paths save real time. This skill ensures both get documented before context is lost. </purpose>
When To Activate
<triggers> After completing significant work:
- Multi-step implementations (3+ distinct phases)
- Debugging sessions that found the root cause
- Experiments or spikes that reached a conclusion
- Tasks where you tried multiple approaches
User signals completion:
- "That worked"
- "Done" / "Finished" / "Ship it"
- "Can you summarize what we did?"
- Explicit request for retrospective
</triggers>
Instructions
Step 1: Capture Failed Attempts First
Document what DIDN'T work before it fades from memory:
Be specific about WHY it failed. "Didn't work" is useless. "Race condition between auth callback and state update" is useful.
Step 2: Document The Working Solution
Step 3: Extract Learnings
Step 4: Create Artifact (Optional)
If the learnings are significant, offer to create:
- A skill (if pattern will recur) → invoke
learn-from-this - A snippet (if code is reusable)
- A doc (if context is project-specific)
Output Format
Examples
Example: Debugging Session
Example: Feature Implementation
NEVER
- Skip the Failed Attempts section (it's the most valuable part)
- Write vague failure reasons ("didn't work", "broken", "issues")
- Document only successes (survivorship bias)
- Wait until the next session (context will be lost)
ALWAYS
- Document failures FIRST while memory is fresh
- Include time estimates (helps calibrate future estimates)
- Extract the key insight (the "aha" moment)
- Offer to create skills from significant patterns
- Be honest about what took longer than expected
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/retrospective/SKILL.md
Popularity
- Repository stars
- 55
- Repository forks
- 2
- 7-day trend score
- 20
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 13 May 2026, 6 months ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/elliotjlt/claude-skill-potions/retrospective)