agentleFS
Sign inStart free
sergeeeysergeeey/Claude-cod-top-2026/skills/extensions/session-retrospective/SKILL.md

session-retrospective

Skill9 starsChanged 6 months agoMITView on GitHub

After any meaningful work session — sprint, debugging marathon, feature implementation. Takes 2-3 minutes, saves 10-15 minutes of re-orienting next session. Trigger with: /retro or retro: prefix on any prompt, or explicitly ask "wrap up session". Read activeContext.md auto-commit log section. Check git log --oneline -10 for actual commits made. Never invent commits — [VERIFIED] only. Run to get concrete numbers: Present as: N commits | M files changed | K tests Only use metrics you can [VERIFIED] — skip…

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 7 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    After any meaningful work session — sprint, debugging marathon, feature implementation.

    line 24
  2. Very specific:

    Read activeContext.md auto-commit log section.

    line 33
  3. Distinctive:

    Run to get concrete numbers:

    line 38
  4. Distinctive:

    Only use metrics you can [VERIFIED] — skip if command fails.

    line 47
  5. Very specific:

    If significant patterns found → remind user to update ~/.claude/memory/patterns.md.

    line 70
  6. Distinctive:

    After retrospective, optionally update:

    line 97
  7. Distinctive:

    Do NOT invent commits — read git log to verify actual work done

    line 108
  8. Distinctive:

    If session had 0 commits → still useful to capture decisions and blockers

    line 109
  9. Distinctive:

    Keep output ≤ 30 lines — this is a summary, not a transcript

    line 110

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- BSV — Brief Skill View | поиск: BSV Скил : session-retrospective TL;DR : Ретро сессии: решения / сработало / нет / дальше Вызов : `/retro`, `retro:` НЕ для : Mid-session, живое кодирование -->

name: session-retrospective description: > [STATUS: review] [CONFIDENCE: high] [REVIEWED: 2026-04-09] Post-session analysis skill. Reviews decisions made, what worked, what didn't, and captures next steps into activeContext.md. Triggers: /retro, retro:, session review, what did we do, wrap up session, session summary, end of session. effort: minimal tokens: ~800

Session Retrospective

When to Use

After any meaningful work session — sprint, debugging marathon, feature implementation. Takes 2-3 minutes, saves 10-15 minutes of re-orienting next session.

Trigger with: /retro or retro: prefix on any prompt, or explicitly ask "wrap up session".

Workflow

Step 1 — Collect session data

Read activeContext.md auto-commit log section. Check git log --oneline -10 for actual commits made. Never invent commits — [VERIFIED] only.

Step 1.5 — Collect git metrics

Run to get concrete numbers:

git log --oneline --since="8 hours ago" 2>/dev/null | wc -l
git diff --name-only HEAD~5..HEAD 2>/dev/null | wc -l
pytest tests/ -q --no-header 2>/dev/null | tail -2

Present as: N commits | M files changed | K tests Only use metrics you can [VERIFIED] — skip if command fails.

Step 2 — Answer 4 questions

What was decided? List architectural/design decisions made this session. Tag each with [VERIFIED] if code was written, [INFERRED] if discussed only.

What worked? Patterns, approaches, or tools that delivered results faster than expected. Tag for patterns.md with [REPEAT].

What didn't work? Approaches that failed, dead ends, wrong assumptions. Tag for patterns.md with [AVOID].

What's next? Top 3 unfinished items or natural follow-ups. These become the next session's ## Current Focus.

Step 3 — Write output

Present retrospective in the Output Format below. If significant patterns found → remind user to update ~/.claude/memory/patterns.md. Offer to update activeContext.md ## Current Focus with next session's top item.

Output Format

### Session Retrospective — [date]

**Decisions made** [VERIFIED / INFERRED]
- ...

**What worked** → [REPEAT] in patterns.md
- ...

**What didn't** → [AVOID] in patterns.md
- ...

**Next session focus** (top 3)
1. ...
2. ...
3. ...

*[VERIFIED] [N] commits · [M] files changed · [K] tests*

Integration with Memory Protocol

After retrospective, optionally update:

FileWhat to add
activeContext.md → ## Current FocusTop item from "Next session focus"
activeContext.md → ## Auto-commit log## Retrospective [date]: [1-line summary]
~/.claude/memory/patterns.md[REPEAT] and [AVOID] entries with [×N] counter
~/.claude/memory/decisions.mdArchitectural decisions tagged [VERIFIED]

Gotchas

  • Do NOT invent commits — read git log to verify actual work done
  • If session had 0 commits → still useful to capture decisions and blockers
  • Keep output ≤ 30 lines — this is a summary, not a transcript
  • Evidence Policy applies: don't present [INFERRED] decisions as [VERIFIED]
  • One retrospective per session — don't run mid-session, it will be incomplete

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free