leiloeiro-edital

Pass

Audited by Gen Agent Trust Hub on Aug 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (auction notices/editais) as part of its core workflow. While this is the intended purpose, it creates a surface for indirect prompt injection where malicious instructions hidden in a document could influence the agent's behavior.
  • Ingestion points: The skill explicitly instructs the agent to "receber um edital (ou informações dele)" (receive a notice or information from it) in SKILL.md.
  • Boundary markers: The instructions include structured blocks (Blocos 1-8) which help segment data, but there are no explicit instructions for the agent to ignore natural language commands found within the analyzed documents.
  • Capability inventory: The skill includes a scripts/governance.py file capable of local file writes (action_log.jsonl) and utilizes platform tools (claude-code, gemini-cli, etc.).
  • Sanitization: No explicit sanitization or filtering logic is provided for the external text input.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 2, 2026, 10:05 PM
Security Audit — agent-trust-hub — leiloeiro-edital