edge-verifier

Pass

Audited by Gen Agent Trust Hub on Mar 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes text from SKILL.md files and user input without delimiters or sanitization.\n
  • Ingestion points: The hint_sweep.py script reads hints from SKILL.md files, and verify_edges.py takes arbitrary text as input.\n
  • Boundary markers: None identified; content is directly interpolated into system and user prompts for the LLM.\n
  • Capability inventory: Both main scripts call the 'memory' skill to write new edges or lessons to the database.\n
  • Sanitization: No evidence of input validation or escaping before LLM processing.\n- [COMMAND_EXECUTION]: Uses subprocess.run to execute the run.sh script of the 'memory' skill for database interactions. This is an expected integration pattern.\n- [EXTERNAL_DOWNLOADS]: Fetches vendor-specific dependencies (graph-memory and scillm) from the author's GitHub repositories.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 17, 2026, 06:35 AM
Security Audit — agent-trust-hub — edge-verifier