repo-first-defense
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes standard command-line tools such as
git,pnpm,find, andgrepto perform repository inventory and security scans. These operations are restricted to auditing the environment and are consistent with the skill's primary function. - [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it is designed to read and process untrusted files within a repository (e.g.,
CLAUDE.md,.cursorrules, and agent instructions). However, it specifically identifies this as a threat and instructs the agent to treat such content as untrusted data rather than authoritative commands, applying a defensive posture. - [EXTERNAL_DOWNLOADS]: The skill recommends the use of well-known and trusted security tools, including Google's
osv-scannerand the officialgitleaksGitHub Action. These tools are cited as part of a recommended security workflow and do not involve the execution of untrusted or unknown remote code.
Audit Metadata