anthropic-validator

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The validator instructs the agent to ingest full asset contents into reviewer contexts and emit detailed YAML findings (including "violation" text) but gives no redaction guidance, so any secrets present in assets may be exposed verbatim in generated reports.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). This skill spawns a documentation-fetching agent at runtime that retrieves and injects official-doc pages (e.g., https://docs.anthropic.com/en/docs/claude-code/skills) into the reviewer prompts to control validation behavior, which is a runtime external fetch that directly influences agent instructions.