Your Skill Name

Your Skill Title

When to Use

• Use when the task is clearly one bounded job
• Use when the inputs and outputs are easy to define up front
• Use when the workflow is stable enough to encode as reusable instructions

When Not to Use

• Do not use for neighboring tasks that deserve their own skill
• Do not use when the task is so broad that the skill boundary becomes fuzzy
• Do not use when the workflow is still too unstable to describe precisely

Inputs

• Required input 1
• Required input 2
• Optional local conventions such as AGENTS.md

Output Files

• Primary output
• Optional companion output

Constraints

• Preserve canonical user-authored material unless explicitly told otherwise
• Prefer deterministic, bounded steps
• Mark uncertainty explicitly

Procedure

1. Read the nearest local instructions first.
2. Confirm the task matches this skill's boundary.
3. Gather only the minimum required context.
4. Produce the expected output format.
5. Validate that the result matches the intended job and did not drift into adjacent work.

Definition of Done

• The task stayed within one clear job boundary
• Inputs and outputs are explicit
• The result format is consistent and reusable
• Uncertainty and open questions are surfaced rather than hidden

Notes

• Add references/ only when a reusable template, checklist, or domain guide clearly helps.
• Add scripts/ only when deterministic behavior or external tooling is genuinely needed.
• Add agents/openai.yaml when you want stable UI metadata or explicit invocation policy.