sf-industry-cme-epc-model
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits a vulnerability surface where untrusted user input (such as Product Names or Codes) is interpolated into shell commands and metadata files without explicit sanitization instructions.
- Ingestion points: User-supplied product names, codes, and attribute values gathered during Phase 1 and 1A in
SKILL.md. - Boundary markers: Absent. The skill instructions do not specify using delimiters or blocks to isolate user-supplied strings from the rest of the prompt or command structure.
- Capability inventory: The skill uses high-privilege capabilities including
sf data query(SOQL execution),sf project retrieve, andsf project deploy(metadata deployment) as documented in the CLI and Validation section ofSKILL.md. - Sanitization: Absent. There are no instructions for the agent to validate or escape characters in the user input before placing them into the
WHEREclause of a SOQL query or within JSON DataPack files.
Audit Metadata