Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, well-sequenced closure protocol with explicit gates, validation calls, and well-signaled one-level reference files. Its primary weakness is verbosity: several sections restate the workflow and closure rules and could be consolidated.
Suggestions
Merge the overlapping 'Stage purpose', 'Use when/Do not use when', and 'Finalization-quality rules' content into the Workflow and Preconditions sections to remove redundancy.
Move the long paper-bundle manifest inventory list into a reference file and keep only the gating checks inline.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~400-line body is domain-specific operational guidance rather than generic concepts Claude already knows, but sections like 'Stage purpose', 'Use when', 'Do not use when', and 'Finalization-quality rules' overlap materially with the Workflow and Preconditions sections and could be tightened. | 2 / 3 |
Actionability | It gives concrete, executable guidance: specific artifact calls (e.g. artifact.validate_manuscript_coverage(detail='full'), artifact.render_git_graph()), exact manifest paths (paper/paper_bundle_manifest.json), and explicit claim classifications and resume-packet questions. | 3 / 3 |
Workflow Clarity | A clear 7-step Workflow is sequenced with an explicit Preconditions/gate section, validation checkpoints (paper-state validation calls), and a feedback loop that routes premature finalization back through the decision stage. | 3 / 3 |
Progressive Disclosure | The body is well-sectioned and pushes denser material to three real, one-level-deep reference files (finalization-checklist.md, resume-packet-template.md, checkpoint-memory-template.md) that are clearly signaled in an 'Extra references' section. | 3 / 3 |
Total | 11 / 12 Passed |