AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Wrap Up

skill-ccomkhj-skills-wrap-up · by ccomkhj

Phase 4 of long-haul — the finishing move. Fires once the /goal condition is met (the goal auto-clears) or the round cap trips. Reports goal-vs-achieved with the final check output, the explore/exploit path each round took, and the net kept diff; then commits the incumbent on a feature branch and opens a PR. Always confirms the branch and PR body with the user before pushing. Use as the last phas…

No reviews yet
0 installs
10 views
0.0% view→install

Install

$ agentstack add skill-ccomkhj-skills-wrap-up

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-ccomkhj-skills-wrap-up)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Wrap Up? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

wrap-up

Phase 4 of long-haul. Runs when the haul ends — the goal cleared, or the round cap was hit. It does two things: report what happened, then ship the incumbent. The PR is outward-facing — confirm with the user before pushing.

Report — write SUMMARY.md

Read STATE.md (round log + ## Tried), every R.md, the current incumbent, and — if the haul ran detached — .longhaul/PROGRESS.log (the per-loop trail an overnight run leaves instead of chat). Write .longhaul/SUMMARY.md (template in [../long-haul/reference/file-formats.md](../long-haul/reference/file-formats.md)) covering:

  1. Goal vs achieved — restate the /goal condition and whether it was met. If partial (cap hit), give the incumbent's final check output and how far short.
  2. The path — per round, the mode (explore/exploit) and one-line result, plus which explores were tried and rejected (## Tried). This shows where the win came from — a refinement or a fresh approach.
  3. Net change kept — the cumulative diff of the incumbent against BASE (git diff .. --stat), and the before→after on the success signal.
  4. Correctness — re-run the goal's stated check on the incumbent now and paste the output. Don't claim "passing" from memory.
  5. Follow-ups — anything out of scope, deferred, or newly surfaced.

Evidence before assertion: run the check before writing "goal achieved". An honest partial beats an overstated win.

Ship — branch, confirm, PR

All git/gh operations run in REPO (from STATE.md) — the deliverable's repo, which may not be the cwd the haul was invoked from. The PR opens against REPO's remote, not cwd's.

  1. Branch. The kept work is the longhaul/incumbent branch (its commits, from BASE). Rename it to a readable feature branch for the PR — git branch -m longhaul/incumbent longhaul/ — never open a PR from the default branch. Delete any leftover longhaul/attempt-* scratch branches and git worktree prune so the PR carries only the kept work, not attempt scaffolding. The PR's base is BASE's branch.
  2. Confirm — the gate. Show the user: the branch name, the commit list (git log ..HEAD --oneline), and the proposed PR title + body (from SUMMARY.md). If GATES sets confirm_pr=required (the default), set STATUS: WAITING-USER: confirm-pr and wait for an explicit yes. If confirm_pr=auto (an accepted unattended finish), skip the wait, open the PR as a draft, and never push to a protected or default branch.
  3. Push + PR. git push -u origin , then gh pr create --title "…" --body "…" (add --draft under confirm_pr=auto). Report the PR URL — don't claim it's open until gh returns one.
  4. Finish. Set STATUS: DONE, PHASE: wrap in STATE.md. Tear down any stray worktree (git worktree prune). Ask whether to delete .longhaul/ now or keep it (it's gitignored either way). If /goal is somehow still active, remind the user it auto-clears on success, or to /goal clear.

PR body shape

Pull straight from SUMMARY.md. End the body with:

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Commits you author here end with:

Co-Authored-By: Claude Opus 4.8 (1M context) 

Don't

  • Don't push or open a PR before the user confirms — it's outward-facing and hard to reverse.
  • Don't include .longhaul/ or longhaul/* scratch branches in the PR. They're transient.
  • Don't overstate a partial result. If the cap tripped short of the goal, say so with the final numbers.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.