scan-codebase-health

Pass

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from source files, documentation, and configuration files to perform health checks, which could allow malicious instructions in the codebase to influence agent behavior.
  • Ingestion points: .claude/**/*.md, AGENTS.md, CLAUDE.md, docs/project-config.json, and various source files (.ts, .cs, .py, etc.) across the project.
  • Boundary markers: None mentioned for separating untrusted codebase data from instructions.
  • Capability inventory: File writing to tmp/reports/, spawning sub-agents via spawn_agent, and execution of local Python scripts.
  • Sanitization: No explicit sanitization or validation processes are defined for the data extracted from the codebase before processing.
  • [COMMAND_EXECUTION]: The skill involves running a local Python script (.claude/scripts/code_graph) with arguments (symbols and filenames) derived directly from scanning the codebase. This is a functional requirement but represents a potential injection vector if symbols are maliciously crafted.
  • [DYNAMIC_EXECUTION]: The skill uses spawn_agent to create fresh sub-agent instances for the 'Fresh-Eyes Review' phase, which is a complex multi-agent workflow that expands the execution context.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 5, 2026, 06:42 PM