skills/ruvnet/ruflo/witness/Gen Agent Trust Hub

witness

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the @noble/ed25519 package from the official NPM registry to perform cryptographic signing operations. This is a well-known and trusted library.
  • [COMMAND_EXECUTION]: The skill utilizes the node runtime to execute localized scripts (init.mjs, regen.mjs, verify.mjs, history.mjs) for manifest initialization, signing, and verification. These operations are core to the skill's documented functionality.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes project files and JSON manifests to verify the presence of specific fix markers.
  • Ingestion points: Reads local project source files and manifest files (witness-fixes.json, verification.md.json).
  • Boundary markers: Standard file formats (JSON, source code) are used.
  • Capability inventory: Uses Bash, Read, Write, and Edit tools.
  • Sanitization: The skill performs sha256 hashing and deterministic substring matching, which are robust methods for regression tracking and do not involve interpreting processed data as executable instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 11:38 AM
Security Audit — agent-trust-hub — witness