Content
10%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.
The body is a persona-driven capability catalog: long lists of topics, tools, and traits that Claude already knows, with no executable commands, examples, or concrete review procedures. Its only progressive-disclosure pointer references a nonexistent file, and the workflow is a sequence of vague attitudes rather than verifiable steps.
Suggestions
Cut the capability/behavior/knowledge catalogs to a short overview and move any retained detail into real reference files (e.g., references/security-checklist.md) that actually exist in the bundle.
Add concrete, executable guidance: example review commands (semgrep, codeql, npm audit), a severity-ranked finding format, and a sample review comment showing the expected output style.
Rewrite 'Response Approach' as a verifiable workflow with checkpoints — e.g., scan with tool X, verify findings against the diff, classify by severity, and re-validate suggested fixes.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~170-line body is heavily padded with capability catalogs enumerating knowledge Claude already has (OWASP Top 10, SOLID, N+1 detection, SonarQube, cyclomatic complexity), plus 'Behavioral Traits' and a 'Knowledge Base' section that add almost no novel information. Fits the anchor 'severely verbose; heavily padded'; not a 2 because there is no section that genuinely earns its tokens — nearly the entire body is restated common knowledge. | 1 / 5 |
Actionability | There is no concrete code, command, checklist, or example anywhere — the 'Instructions' section is generic ('Apply relevant best practices and validate outcomes') and 'Response Approach' only describes attitudes. Fits the anchor 'entirely vague or abstract; only describes rather than instructs'; not a 2 because even high-level executable steps (e.g., 'run semgrep --config p/owasp') are missing. | 1 / 5 |
Workflow Clarity | The 'Response Approach' section provides a rough 10-step sequence ('Analyze code context', 'Apply automated tools', ... 'Follow up'), which is a recognizable order of operations, but each step is vague and there are no validation checkpoints or commands. Fits 'rough sequence present but many gaps; steps poorly defined; validation absent'; not a 3 because the steps give no concrete method for accomplishing or verifying any phase. | 2 / 5 |
Progressive Disclosure | Tens of sections of capability listings that clearly belong in separate reference files are fully inlined, and the single pointer — 'open resources/implementation-playbook.md' — targets a file that does not exist in the bundle (no references/, scripts/, or resources/ directories). Fits 'content that clearly belongs in separate files is inlined'; not a 1 because section headers do exist and one disclosure pointer is attempted, however broken. | 2 / 5 |
Total | 6 / 20 Passed |