codeql
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users on how to download the CodeQL CLI bundle from official GitHub releases (
github.com/github/codeql-action). This is a trusted source for development tooling and is necessary for the skill's stated purpose. - [INDIRECT_PROMPT_INJECTION]: The skill provides guidance for setting up workflows that analyze source code, which is untrusted data. It also mentions GitHub Copilot Autofix, which uses an LLM to process results. This creates a surface for indirect prompt injection if a codebase contains malicious patterns designed to influence the LLM. However, this risk is inherent to the primary purpose of code analysis tools and is addressed through platform guardrails and human review processes.
- Ingestion points: External source code analyzed by the configured CodeQL workflows.
- Boundary markers: Standard GitHub Actions environment isolation.
- Capability inventory: File system access for source code, execution of build commands (make, cmake, etc.), and network access for result uploads.
- Sanitization: CodeQL performs static analysis rather than direct instruction execution; Autofix suggestions require human review before application.
Audit Metadata