skill-slug

name	[skill-slug]
description	[one sentence describing what this skill enables Cowork to do]

When to use

[One sentence — the primary scenario where this skill applies.] [One sentence — a second situation or variation where it also applies.] [One sentence — an edge condition or boundary: when a lighter approach is more appropriate.]

Triggers

• [A direct user phrase that reliably signals this skill, e.g. "user says X"]
• [A task-type signal, e.g. "user shares a [document type] and asks for [output type]"]
• [A follow-up signal, e.g. "user asks to refine or expand an earlier output"]
• [A context signal, e.g. "a specific file type or folder is mentioned in the request"]
• [An implicit signal — what the user is trying to accomplish without naming the skill]

Instructions

1. [First step — what Cowork reads or gathers before starting]
2. [Second step — the core processing or analysis action]
3. [Third step — how Cowork structures or organises the output]
4. [Fourth step — quality check or refinement before delivering]
5. [Fifth step — final delivery action or follow-up offer to the user]

Output format

[Opening line — the top-level structure, e.g. "Numbered list of [N] items."]

Each item follows this shape:

• [Field 1 label]: [description of what goes here and its length target]
• [Field 2 label]: [description of what goes here and its length target]

[One line — any count constraint, e.g. "Aim for [N]–[M] items unless the user specifies otherwise."] [One line — closing action Cowork takes after delivering, e.g. "End with a note on which items seem most important."]

Quality criteria

• [Criterion 1 — a structural check, e.g. "Output contains the required [element]"]
• [Criterion 2 — a content depth check, e.g. "Each [item] includes [required detail]"]
• [Criterion 3 — a scope check, e.g. "Output stays within [count/length] target"]
• [Criterion 4 — a user-value check, e.g. "Output would be directly usable without editing"]
• [Criterion 5 — an edge-case check, optional; remove this bullet if 4 criteria are sufficient]

Anti-patterns

• [Anti-pattern 1 — a common structural mistake in this skill's output domain]
• [Anti-pattern 2 — a scope or depth error, e.g. "surface-level output that doesn't …"]
• [Anti-pattern 3 — a voice or format error, e.g. "treating all [input types] as equivalent when …"]
• [Anti-pattern 4 — a process error, e.g. "starting without reading [required input]"]
• [Anti-pattern 5 — optional; remove if 4 anti-patterns are sufficient]

Example

Input:

[Paste a real input here — e.g. a snippet of text, a user request, or a file excerpt that represents the most common activation scenario for this skill. Keep it to 3–8 lines so the example remains scannable.]

Output:

[Paste the ideal output here — the response Cowork should produce for the input above. Match the structure in §Output format exactly. Include all required fields/sections. Aim for a complete output, not a "..." truncated sketch — reviewers need to see the real shape, not a placeholder for it. Remove this parenthetical when you paste the actual example.]

Writing-profile integration

[One sentence — when this skill reads context/writing-profile.md, e.g. "When output exceeds 100 words, Cowork consults context/writing-profile.md and applies the user's stated tone, sentence-length preferences, and pet peeves."] [One concrete rule, e.g. "If no writing profile exists, default to clear and direct prose." — or "This skill produces structured data under 100 words; writing profile is not consulted."]

Example prompts

• "[A direct invocation phrase — the most common way a user requests this skill]"
• "[A variation with a specific input — e.g. a file reference or a scope qualifier]"
• "[A follow-up or edge-case invocation — a less obvious but valid request for this skill]"