schedule-meeting
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from Salesforce records (Accounts, Contacts, Opportunities) to generate calendar invites and email drafts, creating a potential surface for indirect prompt injection if those records contain malicious instructions.
- Ingestion points: External CRM data is ingested via
dispatch_readonlycalls in the resolution steps ofSKILL.md. - Boundary markers: The instructions do not define explicit delimiters for external data, but they provide strict behavioral guidelines for the agent to follow.
- Capability inventory: The skill can create Salesforce
Eventrecords and initiate calendar or email actions through separate connector tools. - Sanitization: No explicit sanitization of field data is mentioned; however, the skill mandates human review and explicit approval for all outputs derived from this data.
- [SAFE]: The skill implements strict confirmation gates, requiring an 'explicit-yes' from the user before sending any calendar invites, drafting emails, or writing
Eventlogs to Salesforce. This ensures a human-in-the-loop review for all actions that involve external communication or data modification. - [SAFE]: All network operations are performed through authorized tools targeting standard Salesforce API endpoints (e.g.,
/services/data/v65.0/graphql), which is the intended behavior for the skill's purpose.
Audit Metadata