crm-sync

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script crm-sync.py to perform status checks, deduplication, and sync management operations.
  • [DATA_EXFILTRATION]: The skill reads sensitive brand profiles, guidelines, and SOPs from ~/.claude-marketing/ and transmits this information to external CRM platforms via the Model Context Protocol (MCP).
  • [PROMPT_INJECTION]: The skill ingests untrusted data from external sources which could contain malicious instructions designed to influence the agent or the target CRM.
  • Ingestion points: Data is loaded from CSV file paths, JSON arrays, manual entries, and other connected MCP platforms (Google Sheets, email platforms, etc.) as described in the 'Input Required' and 'Process' sections.
  • Boundary markers: No explicit boundary markers or isolation instructions are defined to separate untrusted data from the system prompt or agent instructions.
  • Capability inventory: The skill has the ability to execute crm-sync.py, read/write files in the ~/.claude-marketing/ directory, and perform network operations via CRM MCP integrations.
  • Sanitization: The process includes data validation for email and phone formats, but lacks specific sanitization or filtering to prevent prompt injection attacks within the data payloads.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 09:40 AM
Security Audit — agent-trust-hub — crm-sync