lindy-incident-runbook
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a documentation-oriented runbook for incident response procedures and does not contain malicious code or unauthorized data access patterns.
- [COMMAND_EXECUTION]: The skill includes
curlcommands to check the status of Lindy AI services and user-defined endpoints. These commands target official vendor domains or use placeholders, and authenticate using environment variables rather than hardcoded secrets. - [INDIRECT_PROMPT_INJECTION]: The skill instructions involve reading agent task history and integration error logs to diagnose failures, which introduces an indirect prompt injection surface. 1. Ingestion points: Lindy workspace task history, agent task logs, and integration error evidence. 2. Boundary markers: None explicitly defined to separate log content from agent instructions. 3. Capability inventory: Bash(curl:*), Read, Write, Edit. 4. Sanitization: The skill advises preserving sanitized task IDs and error evidence but does not provide specific implementation logic for sanitization.
Audit Metadata