Session Wrapup
End-of-session checkpoint — append diary, update active sprint doc with interim results/retro, write tech-lead handoff, commit and push. Use before /compact or at end-of-day. Distinct from sprint-wrap-up which closes a sprint.
Get the complete skill folder, including its scripts and reference files.
When to use this skill
Use Session Wrapup when an AI agent needs a reusable SKILL.md workflow for this job: End-of-session checkpoint — append diary, update active sprint doc with interim results/retro, write tech-lead handoff, commit and push. Use before /compact or at end-of-day. Distinct from sprint-wrap-up which closes a sprint.
When not to use it
Skip Session Wrapup when the task is outside the coding category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.
How to install
Where will you use it?
mkdir -p ~/.claude/skills/session-wrapup # Save SKILL.md to ~/.claude/skills/session-wrapup/SKILL.md
- Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/session-wrapup/SKILL.md. For a shared project, commit the folder at .claude/skills/session-wrapup/ instead.
- Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
- For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.
Full install guide for Claude, Cursor, and Codex →
Try a starter prompt
After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.
Use the Session Wrapup skill for this task: At end-of-day / session end (user signals "we're done for today"). Confirm any required inputs or files before making changes.
What supports this recommendation?
82/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.
- Documentation
- 92/100
- Practical detail
- 92/100
- Source evidence
- 58/100
- Source signals
- 72/100
Model compatibility evidence
- claudeinferred
The skill text mentions Claude or a closely associated term. This is keyword-derived compatibility; no hands-on model test is recorded.
- chatgptuntested
No model-specific signal or recorded compatibility test was found.
- geminiuntested
No model-specific signal or recorded compatibility test was found.
- copilotuntested
No model-specific signal or recorded compatibility test was found.
- llamauntested
No model-specific signal or recorded compatibility test was found.
- perplexityuntested
No model-specific signal or recorded compatibility test was found.
- mistraluntested
No model-specific signal or recorded compatibility test was found.
- grokuntested
No model-specific signal or recorded compatibility test was found.
Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.
What this skill does
# Session Wrap-Up
Persist all session-level learnings and state to disk BEFORE running `/compact` or ending a session. Ensures nothing is discarded with the conversation.
## When to use
- At end-of-day / session end (user signals "we're done for today") - Before running `/compact` (especially mid-sprint) - After a long merge wave or planning session - When approaching a token budget threshold (≥40% weekly)
**Distinct from `sprint-wrap-up`** which is for closing a sprint with retrospective + tagging. This skill is for routine session ends while a sprint is still active.
## Step 1: Verify on main in /workspace
```bash pwd && git branch --show-current ```
Must be `/workspace` on `main`. Abort if not.
## Step 2: Append to plan/diary.md
Add a new `## YYYY-MM-DD HH:MM — <brief title>` section at the bottom with:
- **Pass rate** — start and end numbers if changed - **Merges** — PRs that landed, with net deltas - **Closes** — PRs rejected, with reasons - **New issues filed** — numbers + one-line purpose - **Incidents** — OOM, stuck processes, false-positive regressions, CI glitches - **Context/budget** — tokens burned, triggers, mitigations applied - **Key learnings** — non-obvious findings worth saving (1-sentence each)
Keep it to ~30-60 lines. Bullets, not prose.
## Step 3: Update plan/issues/sprints/N/sprint.md
Fill the `## Results` section (mark as "interim" if sprint still active):
```markdown ## Results (interim — YYYY-MM-DD, sprint still active)
**Baseline progress:** <start> → **<end>** pass / <total> total = **<pct>%** (<+delta>) Sprint goal (<goal>) <met|not yet met>. <gap> tests <remaining>.
### Merged - **#NNN <title>** (<PR>, <+delta> pass) ...
### Closed without merging - **#NNN <reason>** ... ```
And if the sprint has a `## Retrospective` section, append an interim retro with "What went well / What went badly / Process improvements / Key numbers / Sprint-close criteria remaining".
## Step 4: Append the retrospective section in plan/issues/sprints/N/sprint.md
If a retrospective file doesn't exist yet, create one with frontmatter:
```markdown --- sprint: Sprint-N status: interim | final session_end: YYYY-MM-DD ---
# Sprint N Retrospective — INTERIM (sprint still active) ```
Then fill with the same what-went-well / badly / process-improvements sections. Mark "interim" clearly if sprint isn't closed yet.
## Step 5: Update plan/agent-context/tech-lead.md
```markdown --- agent: tech-lead session_end: YYYY-MM-DD next_session_entry_point: read this file, then plan/issues/sprints/N/sprint.md ---
# Tech Lead Context Summary — YYYY-MM-DD
## Baseline state ## Sprint N status ### Merged today ### Closed ### PRs still in flight ### Unassigned ready work (prioritized) ## Team state ## Local state needing push ## Unfinished work / loose ends ## Entry points for next session ```
This file is what the NEXT session reads as its first tool call instead of `claude --resume`. Be thorough — ~100-200 lines of actionable bullets.
## Step 6: Stage and commit
```bash git add plan/diary.md plan/issues/sprints/N/sprint.md plan/agent-context/tech-lead.md git diff --cached --stat git commit -m "docs(session): end-of-session checkpoint — YYYY-MM-DD
✓
Persist session learnings to disk before /compact. - plan/diary.md: <summary> - plan/issues/sprints/N/sprint.md: <summary + retrospective updates> - plan/agent-context/tech-lead.md: handoff for next session
Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>" ```
## Step 7: Push
```bash git pull --rebase origin main 2>&1 | tail -3 git push origin main 2>&1 | tail -3 ```
If the rebase gets stuck mid-pick with "editing commit" state, run `git rebase --continue` — it's a transient git state-machine glitch from empty-after-rebase detection.
## Step 8: Tell the user
Short summary: ``` Pre-compact snapshot committed (<commit-hash>). Pushed. Next session: read plan/agent-context/tech-lead.md, do NOT --resume this session. Safe to /compact now. ```
## Notes
- **Never skip step 2 and 3.** Diary + sprint doc are the persistent project log. Compacting without updating them loses the session's learnings permanently. - **tech-lead.md is NOT optional** — it's the handoff channel that replaces session resume. Write it every time you end a long session. - **Don't wait for sprint close** to fill retrospective sections. Interim retros are fine; the sprint doc carries multiple interim entries over time. - **Interim retro files are allowed** — mark `status: interim` in the frontmatter. When the sprint closes, update to `status: final` and amend.
Intended uses
- At end-of-day / session end (user signals "we're done for today")
- Before running /compact (especially mid-sprint)
- After a long merge wave or planning session
- When approaching a token budget threshold (≥40% weekly)
Implementation notes
- Review the source instructions and adapt inputs before running the workflow.
Try it, then tell us how it went
A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.