linear-plan-build
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Linear issue descriptions, comments, and subissues which could contain adversarial instructions designed to influence the agent's behavior.
- Ingestion points: External data is ingested in SKILL.md (Phase 1) and build-loop.md (Step A) using tools like get_issue and list_comments.
- Boundary markers: While the skill uses an adversarial review process (Step D) and a plan approval checkpoint (Step E), it does not explicitly employ prompt delimiters or instructions to ignore embedded commands within the fetched Linear data.
- Capability inventory: The skill has broad capabilities including file modification (Edit, Write), repository exploration (Glob, Grep), and shell execution for building and testing code (Bash).
- Sanitization: There is no evidence of sanitization or escaping for the ticket content before it is interpolated into the agent's context.
Audit Metadata