codew-release-qa-sweep
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute several bash commands, including
cargo fmt,cargo test,cargo build, andgit diff, to verify the state of the repository. - [REMOTE_CODE_EXECUTION]: The skill executes a project-specific shell script
./scripts/release/check-versions.shand the built binary./target/release/codewhale. This involves executing logic external to the skill's instructions. - [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection due to its ingestion of untrusted repository data while possessing command execution capabilities.
- Ingestion points: The agent ingests output from
git diff,cargo test, and the version check script (SKILL.md). - Boundary markers: None identified in the provided instructions to prevent the agent from following instructions embedded in command output.
- Capability inventory: The skill allows shell execution of
cargo,git, and local project scripts (SKILL.md). - Sanitization: No evidence of output validation or sanitization before the agent processes command results was found.
Audit Metadata