intern
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The workflow involves ingesting change definitions and environment state, creating an attack surface for indirect prompt injection.
- Ingestion points: The skill processes user-provided or codebase changes (Step 1) and evidence from development (Step 3) and production (Step 4) environments.
- Boundary markers: Absent. The instructions do not define delimiters for external data or warn the agent to ignore instructions embedded in that data.
- Capability inventory: The skill orchestrates actions via other tools (
ship,prove, andverify-prod) which likely possess file-writing, code-execution, and production-deployment capabilities. - Sanitization: Absent. No validation or sanitization steps are defined for data retrieved from production or development environments.
- [NO_CODE]: The skill consists entirely of a markdown documentation file (
SKILL.md) containing natural language instructions. It does not include any scripts, tool definitions, or binary executables.
Audit Metadata