he-build
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and act upon external plan data and source code diffs, creating a potential surface for indirect prompt injection attacks.\n
- Ingestion points: The agent is instructed to 'inspect its readiness evidence' from plan files and 'inspects the actual diff' from source code during implementation and integration.\n
- Boundary markers: Absent; there are no instructions to use specific delimiters or to treat the ingested data as untrusted text to prevent the execution of embedded instructions.\n
- Capability inventory: The skill authorizes the agent to perform local implementation (file writes), execute focused checks and tests (command execution), and perform integration tasks (Git operations).\n
- Sanitization: Absent; no logic or instruction is provided for sanitizing, escaping, or validating the content of the external files or diffs before they are processed by the agent.
Audit Metadata