Content
83%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 lean and actionable reference catalog: copy-paste commands, concrete indicators, and a useful severity/defender split. The main weakness is workflow clarity — destructive sequences like the Tomcat WAR deploy lack validation checkpoints, and there is no top-level flow for selecting the right technique for a fingerprinted target.
Suggestions
Add validation steps to destructive sequences, e.g. after the Tomcat WAR deploy check the manager response and hit the deployed path only after confirming deploy succeeded.
Add a short selection flow at the top: fingerprint the server/proxy in front of the target (nginx vs Apache vs IIS vs Tomcat vs Spring) and jump to the matching numbered section.
Move external cross-references (ssrf.md, smuggling.md) into a references/ bundle directory within the skill so the links are verifiable and one level deep.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is almost entirely lean, load-bearing code blocks with terse, high-value comments ('# trailing slash CRITICAL', '# 400 if exists, 404 if not — leaks first chars'). No explanations of concepts Claude already knows and no padding — every token earns its place, matching anchor 5. | 5 / 5 |
Actionability | Copy-paste-ready curl, msfvenom, and wscat commands with concrete endpoints, an explicit default-credential list, exact response-code differentiators, and a working PoC-to-secret-extraction pipeline (heapdump → strings → grep). Fully executable and covers the common cases — anchor 5, not 4, since there are no meaningful gaps. | 5 / 5 |
Workflow Clarity | Each technique has an implicit probe→confirm→exploit micro-sequence (e.g., Tomcat list → default creds → deploy WAR → trigger), but there are no explicit validation checkpoints for the destructive operations (WAR deploy → RCE has no 'verify deploy succeeded' step) and no overall workflow for choosing which technique applies to a given target. Per the guideline capping destructive-operation workflows without validation at 3, this lands at anchor 3 rather than 4. | 3 / 5 |
Progressive Disclosure | Well-organized numbered sections with clearly signaled, one-level-deep delegation ('See skills/exploit/web/smuggling.md') and sensible split-out of severity and defender guidance. It does not reach anchor 5 because no bundle files exist and the cross-references point to paths outside the skill that cannot be verified, but references are clearly signaled and structure is good — anchor 4. | 4 / 5 |
Total | 17 / 20 Passed |