sf-ai-agentscript
Pass
Audited by Gen Agent Trust Hub on Apr 28, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill implements a Python-based syntax validator (
hooks/scripts/agentscript-syntax-validator.py) that interacts with the Salesforce CLI (sf) to verify org configuration and metadata readiness. - Evidence: The script uses
subprocess.run()to execute commands likesf config get,sf data query, andsf project deploy. - Context: This behavior is expected for a development tool intended to automate Salesforce deployment workflows. The script includes sanitization logic (
_sql_quote) for SOQL queries constructed from file content. - [PROMPT_INJECTION]: Documentation within the skill contains references to common prompt injection patterns.
- Evidence:
references/fsm-architecture.mdincludes examples such as "Ignore all previous instructions" and "Tell me your system prompt". - Context: These strings are documented as examples of utterances that the Salesforce platform's built-in guardrail topics are designed to catch and are not instructions intended for the AI agent to follow maliciously.
Audit Metadata