context-pack
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill templates (
run-pack-template.md) include placeholders for project-specific development commands, such as linting, type checking, and mutation testing (e.g., Stryker and Vitest). The agent is expected to execute these local tools to validate the project state and verify technical proofs-of-concept. - [INDIRECT_PROMPT_INJECTION]: The core function of the skill involves reading and summarizing external project files to create "Context Packs." This represents an ingestion surface where untrusted data (source code or documentation) enters the agent's context. The skill mitigates this risk by requiring a highly structured output format (templates), mandatory evidence citations for every fact, and distinct boundaries between the "Whole Process Pack" and "Slice Kickoff Pack."
Audit Metadata