The Agent Skills Directory

[PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection through its handling of external CRM data from sources like Salesforce and HubSpot.
Ingestion points: Records are fetched from external systems using fetchSourceRecords and fetchAllSourceRecords as documented in references/implementation-guide.md.
Boundary markers: The provided implementation logic does not explicitly use delimiters or instructions to ignore potential commands embedded within the processed CRM records.
Capability inventory: The skill's execution environment includes Write access and Bash(curl:*) for network operations, which could be leveraged if malicious data is processed.
Sanitization: The implementation guide includes transform and validation functions (e.g., isValidEmail, normalizeTitle, normalizePhone) to verify and clean data fields before they are migrated to the target system.
[SAFE]: External URLs used for resources point exclusively to official documentation for established services including Apollo.io, Salesforce, and HubSpot.
[SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation patterns were detected in the provided scripts or configuration.

apollo-migration-deep-dive