Skip to content
Back to skills

Done

ASecurity

Use at the boundary only — when about to claim something is ready to merge/deploy/push/publish, when ending a session that changed code, and when the user asks "fatto?", "finito?", "possiamo mergiare?", or "quanto rischio a pushare?". Not due inside a loop iteration (there the map delta is the report).

  • 3 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 22, 2026
ai-agentsdocker

Security analysis

A100/100

Scanned September 22, 2026

npx -y skills add MissingPackage/nightshift --skill done --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Done?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Done
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/missingpackage-done/badge)](https://www.skillsdirectory.com/skills/missingpackage-done)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: done
description: Use at the boundary only — when about to claim something is ready to merge/deploy/push/publish, when ending a session that changed code, and when the user asks "fatto?", "finito?", "possiamo mergiare?", or "quanto rischio a pushare?". Not due inside a loop iteration (there the map delta is the report).
---

# Definition of Done

## Overview

Completion claims in this workspace follow one report shape. It exists because the user had to extract the same facts by hand after every change ("Hai fatto tutti i test? ... Cosa devo controllare io a mano? Come testo le modifiche?" — typed for five months).

**A "done" at the boundary without this report is not done. Fill every section; write "none" rather than omitting one.** Boundary = PR to a shared branch, release, publication, deploy to shared infrastructure, end of a session that changed code. Inside a loop iteration the report is not due: the three-line digest and the map delta are.

## The completion report (required shape)

```
## Done: <one-line what>

**Evidence** — commands actually run, with outcomes:
- unit: <command> → <result>
- integration: <command> → <result>   (requires real services, e.g. Docker Compose)
- lint/format: <command> → <result>
- build/other gates: <...>

**Not verified** — what was NOT checked and why:
- <item — reason>   (write "nothing" only if truly nothing)

**Your manual check** — what the user should verify by hand, and exactly how:
- <step-by-step, copy-pasteable>

**Conventions & docs** — project rules touched by this change:
- <e.g. domain events emitted for every new mutation? docs/ADR updated per project procedure? migrations?> → <status>

**Residue** — what you met on the path and could not fix in this pass (a fix that needs a decision of the four kinds, or exceeds the budget):
- <item → where it is written (map or tracker), or "none">
```

## Rules

- The report is written for the user, who often reads it **on a phone**: plain language, every project term explained in half a line or dropped, no bare sigla, things cited by **title**. Open with the map head line: the objective was at X, is at Y, Z moves it next.
- Evidence means **fresh output from this session**, not "tests passed earlier" or "should pass".
- A gate you skipped, reported honestly under **Not verified**, is acceptable. A gate silently omitted is a false claim.
- Partial completion → say "Partially done", keep the same shape, list the remainder.
- Risk questions ("quanto rischio a pushare?") → answer with this report plus a one-line risk verdict grounded in the Not-verified section.

## Rationalizations

| Excuse | Reality |
|---|---|
| "Small change, report is overkill" | Small changes broke the cluster twice. The report takes 60 seconds. |
| "Tests passed before my last edit" | Then they haven't tested your last edit. Re-run. |
| "The user will test it anyway" | He tests what you TELL him to test. That's the Manual-check section's job. |
| "I'll just say done and details on request" | The requests are exactly the four questions this shape answers. Pre-empt them. |

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.