agentleFS
quatico-solutionsquatico-solutions/agent-skills/skills/bye/SKILL.md

bye

Skill2 starsChanged 4 months agoMITView on GitHub

Use when the user says /bye, \"wrap up\", \"end session\", or similar. Reconstructs full session history including compacted context, creates a sessionlog (if the project has a sessionlog directory), commits changes, and summarizes next steps.

  • Reads credentials
  • Installs packages

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    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:

    If running in a project with a local CLAUDE.md or AGENTS.md, check for a "Session Wrap Up" heading.

    line 17
  2. Distinctive:

    Follow the tool-specific restoration guide:

    line 24
  3. Distinctive:

    Cursor: no restoration procedure yet.

    line 26
  4. Distinctive:

    If restoration finds no prior work beyond current context, proceed — but log that restoration was attempted.

    line 29
  5. Distinctive:

    After restoring history, classify the session:

    line 35
  6. Very specific:

    If skipping, note Sessionlog: Skipped — [reason] in the final summary.

    line 79
  7. Very specific:

    Commit message: follow the project's commit convention (CLAUDE.md / AGENTS.md, or the commit-notation skill) when one is declared, and add …

    line 94

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Session Wrap-up

Project-Specific Instructions

If running in a project with a local CLAUDE.md or AGENTS.md, check for a "Session Wrap Up" heading. If found, follow those additional instructions alongside (and in addition to) the steps below. That section is also where the project declares its sessionlog directory — see sessionlog-template.md.

Restore Full Session History First

Reconstruct full history before anything else: context compaction hides earlier work, and without it the sessionlog will be incomplete.

  1. For long or compacted sessions, delegate session-file analysis to a subagent (see subagent-tasks.md)
  2. Follow the tool-specific restoration guide:
    • Claude Code: claude-code-session-restoration.md
    • Cursor: no restoration procedure yet. Use the current conversation and git log, and note "restoration limited (Cursor)" in the summary.
  3. Combine restored history with current context before continuing

If restoration finds no prior work beyond current context, proceed — but log that restoration was attempted.

Parallel session safety: The user may have multiple sessions running. Filter by $CLAUDE_SESSION_ID and timestamp correlation. Never combine work from other sessions. If uncertain, ASK.

Session Type Detection

After restoring history, classify the session:

SignalTypeAction
"isSidechain": true or in subagents/ dirSubagentSTOP — do not run /bye
messageCount <= 2, first msg contains "Context: This summary"Metadata sessionSKIP — not a real work session
First messages reference executing a plan; recent file in ~/.claude/plans/Plan executionRead plan file; sessionlog documents execution vs plan
System message contains "Plan mode is active"Plan creationPlan file is the deliverable — usually skip sessionlog (see step 3)
None of aboveNormalContinue with checklist

Session Wrap-up Checklist

  1. Determine scope — everything between last /bye (or session start) and now. Verify each item was discussed in THIS conversation.
  2. Assess work — files created, files modified, decisions made, research done, tasks completed, tasks remaining. Then refresh live state before calling anything open or done. The conversation's picture of PRs, builds, tickets and branches is a snapshot from when they were last looked at, and they keep moving while a session sits idle — a wrap-up once listed a PR as "awaiting review and merge" hours after it had been merged. For every item the summary or sessionlog will call pending, done or blocked, check its source now: the PR's state (gh pr view / bb pr view — before the first bb call, run bb --version | grep -q quatico-bb; on a miss, skip bb, mark the item unverified and name the installed bb and brew install quatico-solutions/tap/quatico-bb), the CI status of the latest commit, the ticket's status in the tracker, the branch after git fetch. Drop what is already done; mark what cannot be checked as unverified. Never carry a status forward from memory.
  3. Sessionlog needed? Would anything important be lost if we clear this session now? Commits already capture what changed and when. A sessionlog is only worth creating for context not in the committed artifacts.

    Session produced...Sessionlog?Why
    Plan file onlySkipPlan file captures everything
    Q&A only, no file changesSkipNothing to preserve
    Code/doc changes with clear commitsSkipGit history tells the story
    Feature with decisions, alternatives tried, non-obvious rationaleCreateContext not in the code
    Single file, but research/intent/sources behind itCreateLog captures the why
    Plan execution with deviations or partial completionCreatePlan-vs-reality delta is valuable

    Project rules override this table. A documentation or knowledge-layer repo may require a sessionlog for every session; if ## Session Wrap Up says so, follow it and skip nothing.

    If skipping, note **Sessionlog:** Skipped — [reason] in the final summary.

  4. Sessionlog — find the directory per sessionlog-template.md. If no directory exists, skip this step entirely — do NOT create directories. If one exists, create or update the sessionlog per the template.
  5. Update project status — if work relates to projects/*/, update its status.md.
  6. Handle git — see git decision table below.
  7. Print final summary — see template below.

Git Decision Table

SituationAction
Files I created/edited THIS sessionAuto-commit
Untracked files from beforeASK
Modified files I didn't touchASK — likely parallel session
.env, credentials, secretsNEVER, warn user

Commit message: follow the project's commit convention (CLAUDE.md / AGENTS.md, or the commit-notation skill) when one is declared, and add Session wrap-up: YYYY-MM-DD as the last body line. Otherwise use [Brief description]\n\nSession wrap-up: YYYY-MM-DD.

Push if remote tracking exists.

Final Summary Template

## Session Complete

**Accomplished:**
- [item 1]
- [item 2]

**Committed:** [hash]
- [file list]

**Pending:**
- [ ] [task 1]

**Session:** Reconstructed from N compaction(s) · ~Xk input / ~Yk output tokens
           ↑ only include if compactions > 0; round tokens to nearest k

**Sessionlog:** `<sessionlog directory>/[file].md`

Ready to clear context.

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.