instantly-core-workflow-a
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes campaign specifications and reviewed audience data which are external inputs.\n
- Ingestion points: The agent reads campaign specifications, consent basis, suppression rules, and lead data from external files as instructed in SKILL.md.\n
- Boundary markers: No explicit structural delimiters or specific instructions to ignore embedded commands are defined for the campaign specification ingestion.\n
- Capability inventory: The skill utilizes
WebFetchfor API interaction, andWrite/Editfor recording workflow outcomes and evidence.\n - Sanitization: The instructions explicitly mandate the redaction of sensitive fields and strictly prohibit writing credentials, lead data, or email content to files.\n- [DATA_EXFILTRATION]: The skill communicates with official Instantly.ai API endpoints at
api.instantly.aiusingWebFetch. This is legitimate operational behavior targeting a well-known service for campaign management.\n- [EXTERNAL_DOWNLOADS]: The skill references official@instantlyai/sdkand@instantlyai/clipackages as implementation resources for interacting with the Instantly platform.
Audit Metadata