Conciseness

Nearly the entire body is executable curl commands and an operator table; the only prose is a single framing line, operational gating ("Once you have confirmed RCE (e.g., `id` or `hostname` output returned)"), and the "Hard rule" budget note. No concept explanations Claude already knows, and every section earns its tokens.

Actionability

Fully copy-paste-ready curl one-liners with parameterized placeholders (<TARGET>, <CALLBACK>) cover the common cases: reflected detection, time-based and OOB blind detection, space/keyword/encoding bypasses, and a complete automated char-by-char extraction loop. The single-call sweep is directly executable after substituting the confirmed sink.

Workflow Clarity

The sequence is clear and gated: Detection → Bypass Techniques → "After RCE Confirmed" extraction sweep → Blind Exfiltration, with an explicit checkpoint ("Once you have confirmed RCE") and inline result checks in the sweep loop. It falls short of a 5 because there is no error-recovery/feedback loop guidance for when detection or a given exfiltration channel fails beyond implying the operator should try the next listed technique.

Progressive Disclosure

Sections are well-organized, there are no nested or orphaned references, and no bundle files exist to mis-navigate. However, the ~110-line body inlines advanced material (the full bypass catalog and the char-by-char oracle automation) that could live in a reference file to keep SKILL.md a leaner overview; the under-50-line exception for a reference-free 5 does not apply.

Specificity

The description enumerates multiple concrete actions with specifics: "exploiting applications that pass user input to OS commands without sanitization", "injection operators (;, |, ||, &&, $(), backticks, newline)", "blind detection (time-based, OOB callback)", and "bypass techniques (space, keyword, encoding)". Coverage is comprehensive across the technique space it claims, matching the anchor for multiple specific concrete actions.

Completeness

The 'what' is clear and specific (exploiting unsanitized user input passed to OS commands, covering operators, blind detection, and bypasses), but the description field contains no 'Use when...' clause or equivalent explicit trigger guidance, capping completeness at 3 per the judging guidelines. The rich trigger list sits in metadata.when_to_use rather than the description.

Trigger Term Quality

It contains the primary natural term "OS Command Injection" plus "OS commands", "bypass", and "time-based", which users would naturally say. Common synonyms and phrasings ("cmdi", "shell injection", "rce", "os.system", "popen") appear only in metadata.when_to_use, not in the description field itself, so a few natural terms are missing.

Distinctiveness Conflict Risk

"OS Command Injection" with operator-level specifics carves a clear niche that is explicitly scoped to OS commands, keeping it distinct from sibling injection skills (SQLi, XSS) and generic RCE skills. Conflict risk is minimal.