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 methodology skill: concrete commands, payloads, decision trees, validation gates, and time-boxed rotation rules throughout, with essentially no wasted tokens. The main structural weakness is that it is a single monolithic document — a master orchestrator of this size would benefit from splitting mindset material and tool tables into one-level-deep reference files.
Suggestions
Split PART 1 (mindset: thinking domains, amateur-vs-pro comparison, anti-patterns) into a one-level-deep reference file (e.g., references/mindset.md) and keep SKILL.md as a routing overview, which would shorten the orchestrator body considerably.
Move the large tool-routing table and per-vuln-class checklists (escalation decision trees, Error-vs-Blind ladder) into reference files per phase or per bug class, keeping only the phase flow and navigation rules inline.
Either bundle the referenced files (docs/auth-sessions.md, bughunter/tools/*) inside the skill's references/ or scripts/ directories, or note that they live in an external shared toolkit, so the paths referenced throughout the body are verifiable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and token-efficient — almost everything is checklists, decision trees, and routing tables with no padding on concepts Claude already knows. Not anchor 5 because of light motivational framing ('Hunting is not "find a bug" -- it is "prove an attack scenario."') and the large volume of detail inlined that assumes less competence than needed. | 4 / 5 |
Actionability | Fully concrete, executable guidance throughout: exact commands ('python3 bughunter/tools/lead_board.py ingest target.com', 'ffuf -ac', 'kxss -> dalfox'), specific payloads (''', '"', '{{7*7}}', '${7*7}', 'SLEEP(10)'), and a complete tool-routing table with justifications. | 5 / 5 |
Workflow Clarity | The 5-phase flow is clearly sequenced with explicit navigation rules for the non-linear cases ('New API endpoint discovered during testing -> Return to Phase 2'), a validation gate ('Run /validate (7-Question Gate)... Any fail? -> KILL the finding'), feedback loops (re-test after fix for incomplete patches), and checklists (session end, pre-report). | 5 / 5 |
Progressive Disclosure | Internal structure is good (PART 1/2/3, clear headers, well-organized tables) and inline references are clearly signaled, but the body is a ~400-line monolith with no one-level-deep reference files: content like the mindset section and the large tool-routing table is inlined rather than split, and referenced paths (docs/auth-sessions.md, bughunter/tools/*) are not present in the skill bundle to verify. | 3 / 5 |
Total | 17 / 20 Passed |