tickets-to-lanes
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from GitHub issues (map tickets and build backlogs) to generate prompts for downstream agent sessions. This creates a surface for indirect prompt injection if the source tickets contain adversarial instructions.
- Ingestion points: The skill reads ticket bodies, comments, and acceptance criteria from GitHub (SKILL.md, Step 2).
- Boundary markers: The skill uses structured logic to isolate specific fields (e.g., 'acceptance criteria', 'spec section') but does not specify explicit delimiters to neutralize embedded instructions.
- Capability inventory: The skill uses
gh apifor issue management. The generated prompts invoke/implement-unattended, which typically has file-system and code-execution capabilities. - Sanitization: No explicit sanitization or filtering of remote ticket content is performed before it is included in the output prompts.
- [COMMAND_EXECUTION]: The skill uses Git and GitHub CLI (
gh) to retrieve project metadata, user identity, and create coordination issues. These operations are restricted to the local repository context and the official GitHub API, matching the skill's stated purpose.
Audit Metadata