DataArt/artisyn-implementation-harness / skills/implementation-harness
implementation-harness
FlaggedInstall the portable AI-dev harness, rebuild its operational graph standalone, harvest project skill improvements into its canonical templates, package direct canonical edits into a portable bundle, or apply canonical updates to an existing project. Use /implementation-harness launch, /implementation-harness reindex-graph, or /implementation-harness patch harvest|prepare|apply.
canonical/skills/dev/patterns/approval-sign-off.md
View on GitHub# Sign-off
A **sign-off** is a stop-and-wait: the developer must explicitly reply before the
lifecycle moves on. Silence, a later unrelated message, or an automated
notification is never a sign-off.
The lifecycle uses sign-offs sparingly. The normal place is the end of a phase
(`patterns/phase-sign-off.md`). Steps inside a phase use intelligent checkpoints
(`patterns/approval-intelligent.md`) and keep going when the content qualifies.
Three shapes:
- **Content sign-off** — the developer reviews what was produced and confirms it.
Show the actual content, in full if short, or a faithful summary for a long
artifact. A summary that hides what changed is not a basis for approval.
- **Manual operation** — something only the developer runs: a commit, a push, a
guarded command. Propose the exact command and wait for them to confirm; then
check the result yourself where you can (Git, a file). Never run it yourself, even
once they've said yes.
- **External confirmation** — a fact only the developer can attest to, such as "I
looked at the UI and it's right". Never infer it or self-certify it.
## Self-run checks
Running an available, non-guarded check yourself (build, typecheck, test, curl)
is routine work. Do it, record it with the metrics wrapper, and report the real
result with `--source self`.
A self-run check **is** enough to pass a step checkpoint when the plan's
verification item is something a command can answer and the command passed. It is
**not** enough for things only a human can see or decide: UI/UX appearance,
behaviour on the developer's machine when it plausibly differs from yours, a
guarded command, or a divergence from the plan. Those stop and ask.
A worker's own report, or a satellite reviewer's `go`, is input to your decision,
never the decision. Verify the claimed change actually landed (`git status`,
`git diff`, real output) before you treat it as done.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |