Content
85%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.
An exceptionally actionable, dense skill body with concrete executable commands, complete rule examples, explicit validation checkpoints, and strong decision-gate routing. Its weaknesses are mild padding (MITRE table, duplicated install commands) and a monolithic single-file layout where the YARA syntax reference and lookup tables would serve better as separate reference files.
Suggestions
Move the YARA string types / condition essentials syntax reference and the tools, signatures, and error tables into references/ files (e.g. references/yara-syntax.md, references/retrohunt-platforms.md), keeping SKILL.md as an overview with clearly signaled links.
Remove the MITRE ATT&CK mapping table and the duplicated YARA-X install lines from Section 7 — they duplicate the Quick Reference without adding actionable value.
Trim the Detection Signatures table to entries not already derivable from Sections 1 and 4.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is almost entirely commands, code, and lookup tables with minimal explanation of concepts Claude already knows, but minor padding exists: the MITRE ATT&CK mapping table, the duplicated 'cargo install yara-x' between Quick Reference and Section 7, and the Detection Signatures table that partially overlaps earlier sections. Efficient overall with a few tokens that could be trimmed. | 4 / 5 |
Actionability | Fully executable throughout: install commands, strings/xxd/pefile extraction pipelines, yarGen invocations with flags, complete VT Retrohunt/LiveHunt and ANY.RUN curl API calls with JSON bodies, complete YARA rule examples, and a concrete problem/cause/solution error table. Copy-paste ready and covers the common cases. | 5 / 5 |
Workflow Clarity | The Decision Gate routes each situation to a numbered section; sample-to-rule is a 5-step pipeline; the yarGen review checklist and the end-to-end retrohunt sequence include explicit validation checkpoints ('must match known sample, must NOT match goodware', 'Expect 0 matches') with feedback loops for false positives and YARA-X rejections. | 5 / 5 |
Progressive Disclosure | No bundle files exist and the ~430-line body is monolithic. Sections are clearly headed and navigable, but reference material that belongs in separate files — the YARA string-type and condition-syntax references, and the tools/signatures/error tables — is inlined in SKILL.md. Fits 'some structure but content that should be separate is inline' better than the minor-gaps anchor. | 3 / 5 |
Total | 17 / 20 Passed |