Content
78%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 highly actionable — complete CLI invocations, options, outputs, configuration, and error codes — with a well-sequenced six-phase architecture and explicit degradation/fallback behavior. Its main weaknesses are dead references (constitution.md, SPEC.md, shared/feature_flags.py are missing from the bundle) and inline architecture detail that should live in the referenced files.
Suggestions
Ship the referenced "constitution.md" and "SPEC.md" in the bundle or remove the references; dead links break navigation and the Constitution section currently summarizes a file the reader cannot open.
Move the detailed per-phase architecture (Phase 1-5 internals, verbose JSON diagnostics) into SPEC.md and keep only the phase list and routing table in SKILL.md, deepening the progressive-disclosure split.
Reorder so setup precedes usage: put the required environment variables and credential-validation behavior before the command reference, forming an explicit configure-then-run workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with skill-specific facts Claude could not know (thresholds, feature flags, CLI path overrides, error codes) with only minor trimmable padding — the "A billion pages, a billion facts" epigraph and the full verbose-JSON sample. It matches anchor 4 (efficient with minor over-explanation) rather than anchor 3, since almost all content is non-inferable configuration detail. | 4 / 5 |
Actionability | Every command ships as a complete, copy-paste-ready invocation ("python3 scripts/encyclopedia.py search \"repo:anthropics/anthology describe auth middleware\""), with exhaustive options, example JSON outputs, env vars with defaults, and error codes. This matches anchor 5 (fully executable commands covering the common cases). | 5 / 5 |
Workflow Clarity | The six-phase pipeline is explicitly sequenced, with a clear degradation hierarchy, circuit-breaker probe/recovery loop, and dry-run/verbose verification modes — matching anchor 4's "clear sequence with most checkpoints present". It falls short of anchor 5 because there is no ordered user-side workflow (environment setup and credential validation appear only after the commands) and no explicit validate-then-proceed checkpoint for the reader. | 4 / 5 |
Progressive Disclosure | The body references "constitution.md", "SPEC.md", and "shared/feature_flags.py", none of which exist in the bundle, while inlining the full phase architecture that SPEC.md is supposed to hold — matching anchor 3 (references present but not cleanly resolvable; content that should be separate is inline). It is not anchor 2 because section structure is real and the referenced scripts (encyclopedia.py, source_health.py, semantic_cache.py, source_profiler.py, cgcli/, anime-mori/) all exist. | 3 / 5 |
Total | 16 / 20 Passed |