instantly-migration-deep-dive
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill incorporates robust security instructions, explicitly prohibiting the hardcoding, printing, or unauthorized transmission of API credentials. It directs the agent to use approved server-side secret managers and maintain endpoint-specific scoping to adhere to the principle of least privilege.
- [EXTERNAL_DOWNLOADS]: The skill references official software development kits and command-line interfaces (@instantlyai/sdk, @instantlyai/cli) along with official documentation and OpenAPI schemas from Instantly's verified developer domains (developer.instantly.ai and api.instantly.ai). These references are appropriate and necessary for the skill's stated purpose of API migration.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data, which presents a theoretical ingestion surface, managed as follows:
- Ingestion points: The skill reads repository source code using Read/Grep and fetches external API documentation via WebFetch.
- Boundary markers: It provides explicit instructions to redact sensitive fields and follow a structured handoff format to maintain context clarity.
- Capability inventory: The skill is granted Write and Edit permissions for code implementation, but these are restricted by strict "Approval Boundaries" that prevent production mutations without explicit owner consent.
- Sanitization: The instructions require the agent to verify bearer keys and validate operations against exact required scopes, failing closed if a schema or tenant mismatch is detected.
Audit Metadata