pr-to-lint-rule
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The pr-to-lint-rule skill processes repository files and user input to generate automated lint rules.
- Ingestion points: The skill analyzes the project's codebase, existing custom rules, and natural language descriptions provided by the user via the $ARGUMENTS parameter (SKILL.md).
- Boundary markers: The skill does not currently implement specific boundary markers or explicit instructions for the agent to ignore instructions embedded within the codebase it analyzes.
- Capability inventory: The skill is capable of updating CLAUDE.md and executing npm test as a verification step (SKILL.md).
- Sanitization: Repository content is read directly without a dedicated sanitization layer before generation.
- Context: As a developer tool for code generation and maintenance, reading and interpreting existing code is the primary and legitimate function of this skill.
- [COMMAND_EXECUTION]: Various components of the project execute shell commands for build, test, and safety verification.
- Evidence: The ship-pr skill uses commands like npm run lint and npm test as quality gates (ship-pr/SKILL.md). The testing harnesses and evaluation scripts also execute these commands to verify agent performance.
- Context: These commands are typical for the development environment the tool is designed to support.
- [SAFE]: The project demonstrates a strong focus on security by providing tools to mitigate risks.
- Evidence: The safe-bash-guard-v2.hook.mjs file provides a robust implementation for intercepting and blocking dangerous tool calls, specifically targeting destructive actions, credential theft, and remote code execution patterns like curl piped to shell interpreters.
Audit Metadata