cursor-local-dev-loop
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to use Cursor's context-gathering features like @Codebase and @Git to plan and implement changes. This creates an attack surface where untrusted data from the local repository could influence the agent. * Ingestion points: Repository files, documentation, and git history accessed via @Codebase, @Docs, and @Git reference markers in SKILL.md. * Boundary markers: The skill does not provide specific instructions to the agent to treat repository content as potentially adversarial or to ignore embedded instructions. * Capability inventory: The skill configuration in SKILL.md allows for Read, Write, Edit, and Bash(cmd:*) operations, providing significant reach if an injection occurs. * Sanitization: There are no mechanisms described for validating or sanitizing the content ingested from the codebase before the agent acts upon it.
Audit Metadata