spec

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

• Third-party content exposure detected (high risk: 0.85). The required workflow can ingest outsider-authored free text when the input is a GitHub issue URL/identifier (SKILL.md:13) and the skill “fetch[es] the issue body and title” (SKILL.md:65), which is then used as upstream input to draft the spec (SKILL.md:73), thereby feeding that outsider text into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches a GitHub issue at runtime (https://github.com///issues/) and uses the issue body/title as the upstream input that directly controls the agent's prompts.