plan-strict

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

• Third-party content exposure detected (high risk: 0.75). The required workflow explicitly reads an upstream input that can be a GitHub issue body or other outsider-authored artifact (e.g., “GitHub issue URL or identifier” → “fetch the issue body and title”), and then uses that free text to draft the plan/index/task files that are fed back into the agent’s LLM context during generation.

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

• Potentially malicious external URL detected (high risk: 0.90). This skill explicitly accepts and fetches GitHub issue inputs (e.g., https://github.com///issues/) at runtime and uses the issue body/title as the upstream input that will directly control the agent's plan generation.