campaign-sending
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEDATA_EXFILTRATIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses PII (emails, names, and company data) from local CSV files to upload them to the Instantly.ai API. This transmission is central to the skill's primary function of email sequencing and uses the service's official API endpoints.
- [CREDENTIALS_UNSAFE]: The skill manages authentication using the
INSTANTLY_API_KEYenvironment variable. The workflow instructions specifically advise the agent to request credentials from the user during the initial setup rather than utilizing hardcoded secrets. - [INDIRECT_PROMPT_INJECTION]: The skill processes lead data from external CSV files which could contain untrusted input.
- Ingestion points: Reads local CSV files in Step 1 and a DNC context file in Step 2 of
SKILL.md. - Boundary markers: None explicitly defined for field mapping and variable interpolation.
- Capability inventory: Network API operations for lead management and campaign creation defined in
references/instantly-api.md. - Sanitization: Includes a dedicated step to filter and remove domains found on a DNC list before processing leads.
Audit Metadata