code-quality-scan
Pass
Audited by Gen Agent Trust Hub on May 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands such as
nix build,nix flake check, andjust lint. Running these commands relies on the integrity of the project's configuration files; if the repository contains malicious instructions inflake.nixorjustfile, the agent will execute them during the scan. - [EXTERNAL_DOWNLOADS]: The instructions require the agent to 'RESEARCH the best golang code duplication finder before you implement it'. This encourages the agent to download and run unverified third-party tools or libraries from external sources.
- [COMMAND_EXECUTION]: The skill directs the agent to modify the project's
justfileby adding a new 'find-duplicates' command and alias, which involves automated modification of local configuration files. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8):
- Ingestion points: Reads and executes logic from
flake.nix,justfile, and other project-specific build configuration files in the working directory. - Boundary markers: None identified; the agent is not instructed to ignore embedded instructions in the build scripts.
- Capability inventory: Full shell command execution via build tools and file-write capabilities (modifying
justfile). - Sanitization: None identified; the skill directly interpolates project-defined build systems into its execution process.
Audit Metadata