Content
61%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 thorough, well-structured workflow with concrete MCP calls and prompt templates, weakened by duplicated polling boilerplate and a missing error-recovery path for iterative batch review jobs.
Suggestions
De-duplicate the 'save the returned jobId and poll mcp__claude-review__review_status with a bounded waitSeconds until done=true' instruction shared by Steps 2 and 3 by extracting it once into a shared 'Polling' subsection.
Add an explicit failure-recovery step for review jobs (e.g., if status stays non-done past a timeout, retry, check the server log, or surface the error) so the batch workflow has a true validate->fix->retry loop.
Replace placeholder-laden code blocks with at least one fully copy-paste-ready example (concrete waitSeconds value, a real prompt skeleton) or clearly label them as templates.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient with clear headers and terse rules, but the cross-family independence framing is repeated in two blockquotes and the 'save jobId and poll ... until done=true' instruction is duplicated verbatim across Steps 2 and 3, plus trailing explanatory commentary that could be trimmed. | 3 / 5 |
Actionability | It provides concrete MCP tool names, an install command, and detailed prompt templates with example text, but the code blocks contain placeholders like '[Full research context + specific questions]' and '/absolute/path/to/file1' and leave the poll 'waitSeconds' value unspecified, so they are not fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear 6-step sequence with poll-until-done checkpoints and explicit convergence criteria exists, but for an iterative batch-review operation there is no failure-recovery loop (what to do if a job never reaches done=true), and the rubric caps batch-operation workflows lacking validation feedback loops at 3. | 3 / 5 |
Progressive Disclosure | The body is well sectioned (Constants, Prerequisites, Workflow, Key Rules, Prompt Templates) and signals one-level-deep references to output-composition.md and review-tracing.md via markdown links, with no bundle files present; the minor gap is that the prompt templates are inlined rather than split into a separate file. | 4 / 5 |
Total | 14 / 20 Passed |