lodestar-audit

Warn

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes lint, typecheck, and test commands specified by the repository being audited in its .agents/lodestar/context.md file. These commands are executed to validate audit findings.
  • [EXTERNAL_DOWNLOADS]: The skill includes automation for installing the fallow static analysis tool through standard Node.js package managers (npm, pnpm, yarn, bun).
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code and rubric configuration from the repository. While it focuses on static analysis, this creates a surface for potential indirect prompt injection. Ingestion points: source code files and context.md. Boundary markers: absent. Capability inventory: spawnSync (git, fallow, node), fs.writeFileSync. Sanitization: absent.
  • [DYNAMIC_EXECUTION]: The skill uses dynamic imports with computed paths to load common modules from the lodestar-setup companion skill located in a neighboring directory.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 10, 2026, 10:01 AM
Security Audit — agent-trust-hub — lodestar-audit