task-initiation — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited task-initiation (Agent Skill) and scored it 96/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 1 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
The user knows the task. They cannot make their body begin it. This is not a planning problem — it's an activation problem. Do not produce a plan, a breakdown, or a list of considerations. Those things make initiation harder, not easier.
Your job is to produce one concrete physical action that takes <30 seconds, requires no decision, and starts the loop.
When the user says any of:
Do not activate when they want a plan or breakdown — route to request-refactor-plan, grill-me, or clarity-council instead.
A council call here adds latency and decision overhead — the opposite of what initiation needs. The user is already drowning in options. Adding personas makes it worse. Always run inline. Always fast.
Ask these in a single AskUserQuestion call. Keep them tight.
If they can't answer #3, you answer it for them — pick the smallest possible thing.
One block. No headers. No preamble.
Do this now (≤30 seconds): [One literal physical action. "Opensrc/foo.ts. Type the heading// step 1: parse input. Save."]
>
When you've done that, come back and say "done".
That's it. Do not list step 2. Do not explain why. Do not encourage. Do not say "you've got this".
When the user comes back and says "done":
clarity-council — speed matters more than depth here~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.