github-code-interpreter
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local system utilities including git, tree, find, and wc via provided Python and shell scripts to analyze repository structure and metadata. These calls use list-based parameter passing which mitigates shell injection risks, and operations are scoped to intended analysis directories.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from third-party GitHub repositories, which creates a surface for indirect prompt injection.
- Ingestion points: Repository files such as README.md, source code, and configuration files are analyzed by the AI and scanned by helper scripts.
- Boundary markers: None identified.
- Capability inventory: Local file system write access in working directories and execution of system inspection tools (git, tree, find, wc).
- Sanitization: No explicit sanitization or filtering of repository content is performed before analysis.
- [SAFE]: No malicious obfuscation, credential exfiltration, or unauthorized persistence mechanisms were detected. Network activity is limited to standard git operations with well-known services (GitHub).
Audit Metadata