| name | retro |
| description | Review recent local repository history and write a bounded engineering retrospective. |
| compatibility | opencode |
| metadata | {"host":"opencode","migration_phase":"12","mode":"report-first"} |
What This Skill Does
This is the first OpenCode-native /retro slice for gstack. It inspects recent local repository history and writes a concise engineering retrospective.
In this v1 slice:
- inspect recent local git history
- summarize what shipped in the selected window
- call out momentum and friction
- note test health signals when visible from local history
- write a durable local report
Do not modify code, docs, or git history in this v1 workflow.
Required Input
This version can run with no extra arguments.
Optional examples:
Retro for the last 7 days
Retro for this branch
Write the report to .gstack/retro/opencode-smoke.md
Execution Contract
- Stay local; do not load unrelated skills
- Do not use background tasks or todo lists for this v1 flow
- Use local git only
- Write the report file before returning the final answer
- If history is too shallow, still write a bounded report explaining the limitation
Output
Write the report under .gstack/retro/.
Default path:
.gstack/retro/retro-{slug}.md
Honor an explicit output path if the request provides one.
Fixed Report Sections
The report must include these sections:
Inputs Reviewed
Shipping Summary
Wins
Friction
Test Health Signals
Next Focus
Workflow
1. Gather Local History
- inspect recent local commits
- inspect the current branch name
- inspect a short diff summary when helpful
2. Summarize Shipping Activity
Describe what kinds of work actually landed in the selected window.
3. Call Out Wins
Identify the strongest signs of progress.
4. Call Out Friction
Name the main sources of churn, repetition, or uncertainty visible in local history.
5. Note Test Health Signals
Use only visible local evidence such as test-related commits, smoke coverage, or validation changes.
6. Recommend The Next Focus
State the single most useful next discipline or improvement area.
Rules
- Keep the output local and evidence-driven
- Prefer observed patterns over motivational language
- Do not invent team/process details you cannot infer locally
- Do not require remote analytics or external services
- Do not modify repository files other than the requested report output