Use when finishing a task, writing a status report, or handing work back. Emit a short receipt with paths, commands actually run, results, and any leftover blocker. Do not claim done from narrative.
Installs into .claude/skills of the current project.
Are you the author of Completion Receipt?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/netvar1337-completion-receipt)
---
name: completion-receipt
description: Use when finishing a task, writing a status report, or handing work back. Emit a short receipt with paths, commands actually run, results, and any leftover blocker. Do not claim done from narrative.
version: 1.0.0
license: MIT
metadata:
package: local-operator-skills
category: execution
author: Admin
triggers:
- done?
- status
- what changed
- completion receipt
- wrap up
---
# Completion Receipt
## Purpose
Close work with evidence. Pair with `verification-before-completion` for the check itself; this skill is the report shape after that check.
## Workflow
1. **Collect the changed surface.** List files created or edited, commands run after the last change, and the requested outcome.
- Done when every claimed change has a path or a system-state description.
2. **Attach fresh proof.** Use only verification from this turn. If it was not run, run it before writing the receipt.
- Done when the receipt quotes an exit code, test count, hash, or equivalent observable.
3. **Write the receipt.** Follow [references/receipt-template.md](references/receipt-template.md). Omit empty sections rather than filling them with hedges.
- Done when a later agent can resume from the leftover blocker line alone.
4. **Separate fact from inference.** Label assumptions. If the outcome is blocked, say blocked and name the layer.
- Done when no sentence implies success without proof.
## Receipt Laws
- Paths over adjectives.
- Commands over intentions.
- One leftover blocker, or none.
- Keep it short enough to paste into a later prompt.
## Completion Gate
- [ ] Changed paths are exact.
- [ ] Proof command and result are present or the work is marked blocked.
- [ ] Unverified statements are labeled.
- [ ] Next action is a single executable step when work remains.
## Common Failure Modes
- **Tour of the process** instead of artifacts.
- **Stale proof** from an earlier turn.
- **Hidden leftovers** buried in prose.
- **Success language** on a partial or blocked run.
