skills/romainsimon/paperasse/notaire/Gen Agent Trust Hub

notaire

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches data from several well-known and official French government and public sector APIs to enrich its legal analysis.
  • Evidence: References to api-adresse.data.gouv.fr, apidf-preprod.cerema.fr, apicarto.ign.fr, recherche-entreprises.api.gouv.fr, and georisques.gouv.fr in SKILL.md and referenced files.
  • [COMMAND_EXECUTION]: The skill instructions include executing a local Python script to process data from external APIs.
  • Evidence: Instructions in SKILL.md to run python scripts/fetch_notaire_data.py with various arguments for geocoding, DVF transaction lookup, and risk assessment.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for untrusted user data, which is necessary for its primary function as a legal co-pilot.
  • Ingestion points: User-provided context in SKILL.md (address, price, family details, property nature).
  • Boundary markers: The skill uses a structured workflow (Facts/Hypotheses/Analysis/Calculations) to maintain clear context, though no explicit delimiters for user input are defined in the MD files.
  • Capability inventory: The skill can execute local scripts and perform network operations to trusted government domains.
  • Sanitization: Not explicitly detailed in the provided documentation files.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 09:49 AM
Security Audit — agent-trust-hub — notaire