skills/jaganpro/sf-skills/sf-data/Gen Agent Trust Hub

sf-data

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill uses official Salesforce CLI tools and standard patterns for Apex development. It includes robust documentation and cleanup strategies, ensuring that data operations are reversible and compliant with Salesforce governor limits.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as SOQL queries, CSV files, and JSON trees, creating an attack surface for indirect prompt injection.
  • Ingestion points: Data enters via user-provided SOQL/Apex or files in the assets/ directory (CSV, JSON, tree structures).
  • Boundary markers: The skill mandates a "describe-first" workflow using sf sobject describe to validate schema constraints and picklist values.
  • Capability inventory: The skill uses sf data for CRUD operations and sf apex run for code execution within the target Salesforce organization.
  • Sanitization: The skill includes Python-based validation logic (hooks/scripts/soql_validator.py) and instructions to filter out non-createable fields and validate picklist values against org metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:48 PM
Security Audit — agent-trust-hub — sf-data