greptile
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the greptile CLI to perform authentication checks and execute code reviews.
- Evidence: SKILL.md lists commands including greptile login, greptile whoami, and greptile review -b main --agent.
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent or user to install the Greptile tool from a public registry.
- Evidence: SKILL.md specifies the installation command npm i -g greptile.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and analyze repository code and branch diffs, which are untrusted external inputs.
- Ingestion points: Branch commits and repository source code (SKILL.md).
- Boundary markers: No explicit boundary markers or safety instructions are defined in the skill to delimit untrusted code content from agent instructions.
- Capability inventory: The skill uses the Greptile CLI to analyze code and generate findings, which are then processed by the agent.
- Sanitization: No explicit sanitization of code content is performed within the skill's logic.
Audit Metadata