clarify — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited clarify (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 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.
Read the user's message and decide: is this request clear enough to act on immediately, or does it contain ambiguity that would lead to wasted work? If ambiguous, ask ONE clarifying question before touching code, files, or tools.
Auto-trigger on every user message. No invocation needed. This runs as a thin lens between the user's message and your next action.
CLEAR REQUEST? → Act immediately. Do not stall.
AMBIGUOUS? → Ask exactly ONE question, then stop and wait.Do NOT turn this into a multi-step interview. ONE question, then wait for the answer.
Scan the user's message for these signals. If any are present, flag it internally:
Your question must be:
"Should authentication use JWT (stateless, simpler) or session tokens with Redis (revocable, more ops)? I'll default to JWT if you don't have a preference."
"You mentioned 'refactoring the user service' — do you mean splitting it into smaller classes, moving to a different package, or changing the API surface? Or all of the above?"
"Is the performance issue in the database query (slow SQL), the API response (large payload), or the frontend render (slow component)? I can investigate the most likely layer first."
"Can you clarify what you mean?" ← too vague "What are your requirements, constraints, timeline, and budget?" ← too many questions "Let me ask you a series of questions to understand this better." ← violates "one question"
Skip the question entirely when:
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.