delegation-brief — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited delegation-brief (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.
Asks the user up to 5 targeted questions and generates a structured task brief — a ready-to-use instruction for a new Cowork session. Solves the problem of vague requests that lead to unexpected results, damaged files, and wasted tokens.
Target audience: new Cowork users, non-technical professionals.
Russian: «бриф», «составь бриф», «помоги делегировать», «структурируй задачу», «создай бриф», «напиши бриф», «хочу делегировать», «как правильно описать задачу», «помоги сформулировать задачу» English: "brief", "delegation brief", "task brief", "help me delegate", "structure my task", "create a brief", "write a brief", "how to describe a task to Claude"
Determine the language of the user's request:
Trigger phrase — no prior input required. All data is collected through the dialogue in Steps 1–2.
Structured text brief in chat with five sections:
Before launching AskUserQuestion, check: does the user's message already contain answers to one or more of the 5 questions?
Action:
Examples:
Use AskUserQuestion to ask the unknown questions from Q1–Q4 simultaneously (after Step 0). Combining questions into one call reduces friction — the user responds at once without waiting for follow-ups.
Ask the following questions:
Q1 — What needs to be done?
Q2 — What does the expected result look like?
Q3 — Which files or folders does the task involve?
Q4 — What must not be touched?
Use AskUserQuestion for the fifth question separately — it sets the quality bar and requires a deliberate answer.
Q5 — What does success look like?
Generate the brief based on the answers. Rules:
| Situation | Action |
|---|---|
| User selected a preset option | Rephrase into a concrete statement |
| User typed custom text (Other) | Use verbatim, without edits |
| Answer "No restrictions" on Q4 | Record: "No restrictions" |
| Answer "don't know", "not sure", "haven't decided" on any question | Record: "Not specified — clarify before starting" |
| No files involved | Write: "New file — clarify path before starting" |
Output the brief strictly in this format:
If the user's language is Russian:
## Бриф задачи
**Задача:** [что нужно сделать — 1-2 предложения]
**Конечный результат:** [формат и вид готового результата]
**Файлы к изменению:** [список или «Не указано»]
**Что нельзя трогать:** [ограничения или «Без ограничений»]
**Критерий успеха:** [как выглядит выполненная задача]If the user's language is English:
## Task Brief
**Task:** [what needs to be done — 1-2 sentences]
**Expected result:** [format and shape of the output]
**Files to modify:** [list or "Not specified"]
**Off-limits:** [restrictions or "No restrictions"]
**Success criteria:** [what a completed task looks like]After the brief, add a tip on a separate line:
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.