oss-launch-kit
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses the official GitHub API (api.github.com) to retrieve public repository information and README files. This functionality is essential for its stated purpose of project analysis and launch orchestration.
- [SAFE]: Secret handling is managed through environment variables (GITHUB_TOKEN), following security best practices. There is no evidence of hardcoded credentials or attempts to access sensitive local files like SSH keys or AWS configurations.
- [SAFE]: The skill's logic is implemented in deterministic Python scripts without the use of dynamic code execution (eval, exec) or suspicious subprocess calls. It utilizes the standard 'requests' library for network operations.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external GitHub README files, which constitutes an indirect prompt injection surface. However, it employs heuristic cleaning (stripping markdown) and its output is intended for human review rather than autonomous execution.
- Ingestion points: Fetches 'readme_text' from GitHub repositories in 'scripts/fetch_repo_context.py'.
- Boundary markers: The output uses structured markdown headers but lacks explicit 'ignore embedded instructions' warnings for processed data.
- Capability inventory: Limited to network GET requests to GitHub and local file writes (launch-kit.md); no execution or system-level modification capabilities are present.
- Sanitization: Uses regex-based cleaning and markdown stripping in 'scripts/build_product_brief.py' to extract factual signals and normalize content before generation.
Audit Metadata