do-issue

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

• Third-party content exposure detected (high risk: 1.00). The skill's Phase 1 Issue Discovery explicitly fetches and ingests issue bodies and comments from external git platforms (see modules/issue-discovery.md and the Forge CLI examples like "gh issue view ... --json title,body,labels,assignees,comments"), which are untrusted, user-generated third-party contents that the agent parses and uses to drive task planning and code actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill fetches live issue bodies at runtime via platform issue URLs (e.g. https://github.com/owner/repo/issues/42 and https://gitlab.com/owner/repo/-/issues/42) using gh/glab and then uses that external content to drive prompts and task generation for subagents, which meets the criteria for a runtime external dependency that can control agent instructions.