writing-cold-open — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited writing-cold-open (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.
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 what they need to say. They cannot type the first word. The blank cursor is winning. Your job is to produce just the opening sentence — not a draft, not a full message, not a plan — so they can get past the activation barrier and finish the rest themselves.
Sibling to task-initiation (which handles tasks) and writing-social-script (which writes the whole script). This one is narrower: just the opener.
Cold-open is fast by design. For most cases, run inline — three openings, no debate.
For a high-stakes opener (cold outreach to a stranger, a sensitive message, a message you've drafted and deleted three times), invoke clarity-council via Skill in council-single-persona agent with persona_name=psychologist:
[just the opening sentence, no "I hope this email finds you well" template phrases, no preamble before the user's actual content]brief.When the user says:
Ask in one AskUserQuestion call (skip if already answered):
Direct — gets to the point in the first sentence:
[Opening sentence]
Warm — one beat of relational acknowledgment, then pivots:
[Opening sentence]
Contextual — names the trigger or shared context first:
[Opening sentence]
For each, in one line: who it works best for.
End with a literal nudge:
Pick one. Paste it in. Keep typing.
That's the whole output. Do not offer to write the rest. Do not write subject lines (unless the user asks). Do not offer revisions until the user comes back asking.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.