Content
82%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A lean, highly actionable skill with executable curl examples, an explicit approval gate for writes, and good error/pagination checkpoints. The main gaps are a missing update-state mutation example and no explicit verification of mutation success before reporting results.
Suggestions
Add an issueUpdate mutation example (e.g. changing state) since updates are advertised but only creates are shown.
Instruct the agent to check the mutation response's `success` field and `errors` before reporting the identifier/url to the user.
If the skill grows (cycles, labels, more entities), move extended query examples to a reference file and signal it clearly from SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every section earns its place: the auth paths, error handling ("tell the user they need to connect Linear and stop"), and curl examples carry no padding, and it delegates composio details to "the composio skill's 'Raw requests' section" instead of re-explaining. This matches the lean anchor; there is no over-explanation to trim, so it is not 4. | 5 / 5 |
Actionability | Copy-paste-ready curl commands cover search, single-issue read, team listing, issue creation, and commenting, matching mostly-executable guidance. It falls short of anchor 5 because updates are promised ("changing state is a write") but no issueUpdate/state-change mutation example is provided — a minor gap in covering common cases. | 4 / 5 |
Workflow Clarity | The flow is clear (find/read → approval gate "summarize what it will create/change ... and ask for approval" → write → "Always report the resulting issue/comment identifier and url"), with checkpoints for pagination ("if pageInfo.hasNextPage is true, re-query") and auth failure (401/400 handling). It is not 5 because there is no explicit instruction to check the mutation response's success/errors payload before reporting, and the update flow has no example — minor validation gaps. | 4 / 5 |
Progressive Disclosure | The body is well-sectioned ("Find / read issues", "Writes require approval") with all content appropriately inline for a single-file skill of this size; nothing clearly belongs in a separate file. It is not 5 because at ~90 lines it exceeds the under-50-line simple-skill exception, and the full operation reference is inlined with no signaling of where deeper material would live if the skill grew. | 4 / 5 |
Total | 17 / 20 Passed |