plugins-copilot/session/skills/session-checkpoint/SKILL.md
Save a checkpoint in the active session to preserve progress across context windows. Triggers after completing a major task, stage, or milestone, or when context window usage is approaching full.
npx skillsauth add st0nefish/claude-toolkit session-checkpointInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
PROACTIVELY invoke this (without being asked) after completing a major task, stage, or milestone, or when context window usage is approaching full.
Commits staged/unstaged changes and posts structured context to the linked issue.
Gather current state:
inspect the current branch, default branch, recent commits, changed files,
and any linked issue directly with git and native host tooling.
Infer what's in progress and what should come next from conversation context. When auto-triggering, do NOT ask — infer from the conversation. Only use AskUserQuestion if explicitly invoked and progress is genuinely unclear.
Compose the checkpoint comment content with these sections:
=== CHECKPOINT ===
Branch: <branch>
Timestamp: <ISO 8601>
=== NEXT STEPS ===
- <what to pick up next>
=== KEY CONTEXT ===
- <decisions, gotchas, important state that shouldn't be lost>
Commit any uncommitted changes:
git add -A
git commit --no-verify -m "checkpoint: <brief description of progress>"
If there are no changes to commit, skip the commit step but still post the issue comment. Only push if the user explicitly asked to publish the checkpoint or has already approved pushing in this session.
If the current branch matches type/NNN-*, post the checkpoint as an issue comment:
gh issue comment <N> with the body from step 3Briefly confirm: checkpoint committed and (if applicable) posted to issue #N. When auto-triggering, keep to one line. When user-invoked, mention /session:session-resume to continue from this point.
development
Start work from your description — explore the codebase and plan
data-ai
Multi-phase, multi-agent feature workflow: spec → plan → refine → divide → execute → review. Invoke when the user escalates a session-start/session-issue flow to orchestration, or asks to run a non-trivial feature (multiple files, design ambiguity, cross-cutting concerns, correctness-critical paths) through the full multi-agent workflow. For small fixes, prefer session-start.
tools
Browse open issues, pick one, and start work on it
tools
Review, clean up, and open a PR to finalize the work