The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from application logs and health insights which could potentially contain malicious instructions.
Ingestion points: Data enters the agent context through the logs and insights tools as described in SKILL.md.
Boundary markers: The skill does not explicitly define delimiters or instructions to ignore commands embedded within log data.
Capability inventory: The skill is restricted to read-only observability and lacks capabilities for file system modification, external network operations, or shell command execution.
Sanitization: No specific methods for sanitizing or escaping the content retrieved from Convex logs are specified in the instructions.

convex-insights