Conciseness

The body is almost entirely executable commands; the only prose is a one-line framing sentence ("Exploits XML parsers that process external entity definitions, enabling local file reading, SSRF, or denial of service") and section comments. It assumes Claude's competence, explains no background concepts, and every block teaches a distinct vector — matching the 5 anchor "lean and efficient" rather than 4's "minor instances of over-explanation".

Actionability

Every section is copy-paste-ready curl/bash: detection probes, the /etc/passwd confirm, a credential-file read loop, SOAP payloads with namespace matching, SVG and XLSX upload construction, a served evil.dtd with a python listener for blind OOB, and URL-encoded variants. Specific examples cover all common XXE cases, matching the 5 anchor; the only imperfection is the bare comment "# 3. Check listener for exfiltrated data", which is incomplete guidance rather than a missing executable step.

Workflow Clarity

Sections sequence logically from Detection ("Read /etc/passwd (confirms XXE)" is an explicit validation checkpoint) through escalating vectors, and "Common Pitfalls" acts as error-recovery guidance ("If entity expansion disabled → try parameter entities"). Not 5: the blind-XXE workflow ends at the stub comment "# 3. Check listener for exfiltrated data" with no verification command, and the file-read loop has no explicit check that reflected entity content actually appeared in the response before iterating.

Progressive Disclosure

No bundle files exist, and the body is cleanly sectioned (Detection, Basic, SOAP, Upload, Blind, Content Types, Pitfalls) with each block needed at exploitation time. Not 5: at ~133 lines with the multi-command XLSX package construction inlined, part of that detail would sit more naturally in a scripts/ helper so SKILL.md stays a lean overview — this is the 4 anchor's "minor organization gaps".

Specificity

The description lists four concrete action areas — "local file reading via XML parsers", "SOAP/WSDL API exploitation", "blind out-of-band exfiltration", "SVG/DOCX/XLSX upload XXE" — which comprehensively covers the standard XXE attack classes. This matches the 5 anchor ("multiple specific concrete actions; comprehensive coverage") rather than 4, since no meaningful capability gap is evident.

Completeness

It explicitly answers both questions: the "what" is the four named exploitation capabilities, and the "when" is the explicit clause "Use for any challenge involving XML processing, SOAP endpoints, WSDL services, or XML-based file upload parsing" with concrete trigger phrases. This is a direct match for the 5 anchor's exemplar pattern, and clearly above 4 where the "when" could be more specific.

Trigger Term Quality

Natural trigger terms are comprehensive: "XXE" and "XML External Entity injection" (acronym plus synonym), "SOAP", "WSDL", "XML parsers", "SVG/DOCX/XLSX upload", and "XML processing" — phrasings users would actually say. Only trivial variants (a literal ".xml" extension, "xml api") are absent, which still fits the comprehensive 5 anchor better than 4's "a few natural terms missing".

Distinctiveness Conflict Risk

The niche itself is sharply defined (XXE injection with named vectors and formats), but the trigger "Use for any challenge involving XML processing" is broad enough to fire for adjacent XML skills (e.g., XPath/XSLT injection or XML validation tasks). That is minor overlap risk with closely related skills — the 4 anchor — rather than the minimal-conflict 5 anchor.