lodestar-architecture

Pass

Audited by Gen Agent Trust Hub on Nov 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill reads and samples content from the repository's source files (such as index.ts and other representative source files) to assess project architecture and responsibilities. This creates an attack surface where malicious instructions hidden in the codebase could influence the agent's behavior during the review process.
  • Ingestion points: index.ts, representative source files, package.json, and other discovery files across the repository.
  • Boundary markers: The skill does not define explicit delimiters or instructions for the agent to ignore embedded commands within the sampled files, increasing the risk that instructions inside the source code could be interpreted as agent commands.
  • Capability inventory: Reads multiple repository files, executes a local script (audit-state.mjs), and writes a single report file to the filesystem.
  • Sanitization: No sanitization or filtering of the sampled source code is performed before the agent processes it.
  • [COMMAND_EXECUTION]: The skill executes a local script using Node.js to check for state freshness: node <lodestar-audit-skill>/scripts/audit-state.mjs check-freshness. While this execution relies on a sibling skill and appears to be a standard tool invocation within the system, executing scripts derived from dynamic paths is a potential vector for local command execution if the path variable is compromised.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 2, 2026, 01:33 PM