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.
The body is an excellent example of a lean instruction-only skill: it gives concrete tool-level guidance for both runtime branches, includes an explicit ingest verification checkpoint, and avoids all padding. The only gaps are minor — no invocation examples for the named tools and no error-recovery guidance if ingestion or data quality fails.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is ~180 words covering a branching tool-selection scenario with zero padding: it never explains what BloodHound, Active Directory, or attack paths are, and every clause maps to a concrete tool name, constraint ('Never mix their data or credentials'), or deprecation note ('The retired in-house bh_ingest_zip and related wrappers are deprecated'). This matches 'lean and efficient; assumes Claude's competence; every token earns its place'. | 5 / 5 |
Actionability | The body names eleven concrete tools (bloodhound_mcp_upload_collection, bloodhound_mcp_data_quality, bhce_status, bhce_ingest_zip, etc.) with preconditions and constraints ('ingests an authorized ZIP under /workspace', 'only for bounded read-only questions'), matching 'mostly executable guidance; concrete code or commands with minor gaps'. It falls short of fully executable because no invocation examples are given (e.g., a sample bounded Cypher query or argument shapes for find_objects). | 4 / 5 |
Workflow Clarity | The sequence is clear for both branches: identify available tools, then hosted (upload, verify terminal 'Complete', run analysis tools) or standalone (bhce_status, bhce_ingest_zip, bhce_cypher). Validation for the batch ingest operation is present ('Verify terminal Complete' plus bloodhound_mcp_data_quality as a post-ingest checkpoint), so the batch-operation cap does not apply; however there is no error-recovery feedback loop for a failed upload or poor data quality, matching 'clear sequence with most checkpoints present; minor validation gaps'. | 4 / 5 |
Progressive Disclosure | The body is under 50 lines with no bundle files to reference, and its paragraph structure (runtime selection, hosted path, standalone path) is well organized, matching the rubric's simple-skill guidance. The one external reference (/skills/standard/ad/bloodhound-query/SKILL.md) is clearly signaled and one level deep ('Load ... only when its detailed procedure is relevant'), matching the well-signaled one-level-deep reference anchor. | 5 / 5 |
Total | 18 / 20 Passed |