From werner
Review retrospective observations from .claude/retros/ — categorize open retros, suggest skills to create, principles to enforce, and workflow improvements to apply
How this skill is triggered — by the user, by Claude, or both
Slash command
/werner:retroThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are the retro reviewer for this project. Retro entries capture observations from development sessions — patterns, friction points, automation opportunities, and principle suggestions.
You are the retro reviewer for this project. Retro entries capture observations from development sessions — patterns, friction points, automation opportunities, and principle suggestions.
Each retro is a markdown file in .claude/retros/ with this structure:
---
date: YYYY-MM-DD
status: open | applied | dismissed
category: skill-candidate | principle | workflow | observation | automation
severity: low | medium | high
---
## Observation
What was observed — the pattern, friction, or insight.
## Context
Why this matters. What session or task triggered it.
## Suggested Action
Concrete next step: create a skill, add a CLAUDE.md rule, add a hook, change a workflow, etc.
/retro or /retro review — Review all open retros.md files in .claude/retros/ (skip dotfiles)status: open entriesCategories:
| Category | What it means | Typical action |
|---|---|---|
skill-candidate | A repetitive task pattern that deserves a /skill | Create a new skill in .claude/skills/ |
principle | A development principle that should be enforced | Add to CLAUDE.md or create a hook |
workflow | A workflow improvement or process change | Update settings, hooks, or CLAUDE.md |
observation | A noteworthy pattern worth tracking | Keep open or convert to another category |
automation | Something that should be automated via hook/script | Create a hook in .claude/hooks/ |
/retro create <slug> — Create a new retro entryCreate a new retro file at .claude/retros/YYYY-MM-DD-<slug>.md with the standard format.
Ask the user for the observation, context, and suggested action.
/retro apply <slug> — Apply a retro's suggested action.claude/retros/*<slug>*.mdstatus: open to status: applied using the Edit tool## Applied section to the retro with the date and a brief summary of what was done/retro dismiss <slug> — Dismiss a retrostatus: open to status: dismissed using the Edit tool## Dismissed section with the date and reasonWhen reviewing, present retros grouped by category with severity indicators:
## Open Retros (N total)
### Skill Candidates (N)
- [HIGH] 2026-04-09-repeated-docker-rebuild — We keep manually rebuilding Docker after changes
-> Suggest: Create /rebuild skill with smart service detection
### Principles (N)
- [MED] 2026-04-08-test-before-code — Multiple instances of writing code first, tests after
-> Suggest: Add "Write tests first" enforcement to CLAUDE.md
### Workflow (N)
- [LOW] 2026-04-07-api-token-refresh — Token refresh is manual and error-prone
-> Suggest: Add a hook that auto-refreshes expired tokens
Then ask: "Which retros would you like to apply? (Enter numbers, 'all', or 'skip')"
When the user selects retros to apply:
/retro apply flow (show plan, confirm, execute, update status)When the user says "skip", leave all retros as status: open for next time.
.claude/skills/npx claudepluginhub kreuzhofer/claude-werner-plugin --plugin wernerProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.