neo-harness

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The generator script processes untrusted project metadata from the target repository without sanitization.\n
  • Ingestion points: The script reads and parses project configuration files such as package.json, pyproject.toml, and Cargo.toml to extract project names and command scripts.\n
  • Boundary markers: There are no delimiters or explicit instructions provided to the agent to ignore potentially malicious embedded content within the extracted strings.\n
  • Capability inventory: The generated AGENTS.md file is intended to serve as the authoritative workflow guide for agents, and the skill possesses file-write capabilities to create this document.\n
  • Sanitization: The script does not perform any validation, escaping, or filtering of values retrieved from the repository configuration before writing them into the generated markdown template.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 08:00 AM
Security Audit — agent-trust-hub — neo-harness