Content
86%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 tight, highly actionable skill body: exact MCP tool calls with parameters, explicit categorization criteria, and a well-gated confirmation step before delegation. The few imperfections are a circular self-reference about parsing oversized results and unspecified details like the team-query value.
Suggestions
Fix the circular reference in the Gotchas: "parse it from the persisted tool-result file (see step 3)" points at step 3, which never describes the parsing — replace with the actual recovery procedure (e.g., which file path and how to extract the issue list from it).
Specify what the `query` argument to `list_teams` should match (team name, key, or slug) so step 1 is fully deterministic.
Add a brief post-delegation verification step (e.g., re-fetch the delegated issues or check the save_issue results) to close the feedback loop on the batch operation.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and efficient throughout: every line is either an actionable step with exact tool names and parameters, or a non-obvious gotcha (Oz is an agent not a user, proxy errors at large limits). No padding and no explanation of concepts Claude already knows. | 5 / 5 |
Actionability | Guidance is mostly copy-paste ready — exact tool names (`mcp__claude_ai_Linear__list_issues`), parameters (`limit: 50`, `includeArchived: false`, `delegate: "Oz"`), and a required comment template. Minor gaps remain: "list_teams with a query" doesn't specify what the query matches, and the gotcha "parse it from the persisted tool-result file (see step 3)" is circular since step 3 never explains that parsing. | 4 / 5 |
Workflow Clarity | The 8-step sequence is clear with an explicit validation checkpoint before the destructive batch operation (step 7: "present the user with a summary table... Do not proceed until the user confirms") and an error-recovery path for oversized results. It falls short of a 5 only because there is no post-delegation verification step and the persisted-file recovery is not concretely actionable. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines, single-purpose, with no need for external references, and is well organized into two clean sections (Workflow, Gotchas) — meeting the rubric's stated condition for a 5 on simple skills. | 5 / 5 |
Total | 18 / 20 Passed |