afcc-setup-coordinate
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill queries and processes untrusted metadata from the target Salesforce org during its audit phase, which is then used to generate a deployment plan.\n
- Ingestion points: The
scripts/audit_org.shscript executes SOQL queries against the Salesforce org to retrieve user names, queue names, routing configurations, and messaging channel details, saving them toorg_profile.json(described in the AUDIT stage ofSKILL.md).\n - Boundary markers: The skill includes an
APPROVEstage that serves as a manual checkpoint, requiring the user to review the proposed changes before any write operations are executed.\n - Capability inventory: The skill possesses extensive capabilities across its scripts and instructions, including file system access (
Read,Write), execution of local shell scripts (scripts/*), and performing network operations/metadata deployments via the Salesforce CLI (sf api request rest,sf project deploy start).\n - Sanitization: While the audit script uses
jqto parse JSON results, the content of these records is directly interpolated into user prompts and plans without explicit filtering for embedded prompt injection patterns.\n- [COMMAND_EXECUTION]: The skill makes heavy use of theBashtool to interact with the Salesforce CLI and run local orchestration scripts.\n - Evidence: The
allowed-toolsheader inSKILL.mdincludes varioussfcommand patterns and local script paths (Bash(scripts/*),Bash(./scripts/*)).\n - Details: Commands are used to query org state, deploy metadata (
sf project deploy start), and procure voice numbers through REST API calls.
Audit Metadata