nemoclaw-maintainer-security-code-review

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt requires the agent to read full file contents and produce detailed findings (file:line, descriptions) about hardcoded secrets and diffs, which very likely leads to verbatim exposure of secret values in the report unless explicitly redacted.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 1.00). The skill explicitly requires network access to clone and checkout arbitrary GitHub repositories/PRs and to "Read Every Changed File and Diff" (see "Prerequisites" and Steps 1–4), which exposes the agent to untrusted, user-generated content from public GitHub that could contain malicious instructions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires and parses GitHub PR/issue URLs (e.g., https://github.com/OWNER/REPO/pull/NUMBER), clones/checks out the repo at runtime, and reads changed files—thereby fetching external code/content that is injected into the agent's context and can directly control its analysis/output.