The Agent Skills Directory

[SAFE]: The skill defines a secure-by-default posture for API design. It explicitly mandates security controls like rate limiting when authentication is disabled and defines clear boundaries for anonymous access.
[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface for processing external data.
Ingestion points: External request bodies enter the system via declarative endpoints defined in *.endpoint.ts files (SKILL.md).
Boundary markers: Data is constrained by Zod schemas located in node_modules/@objectstack/spec/ (references/_index.md).
Capability inventory: The skill can trigger database operations (object_operation) and automation workflows (flow) (SKILL.md).
Sanitization: The runtime explicitly rejects transform functions in mappings to prevent arbitrary code execution during data processing (SKILL.md).
[COMMAND_EXECUTION]: The skill references local CLI commands os validate and os build for developer-time validation of schemas and predicates (SKILL.md). These are standard tools for the ObjectStack environment.

objectstack-api