The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes npx tsc --noEmit to verify that the hardening changes do not introduce type errors. This is a standard development tool and practice.
[INDIRECT_PROMPT_INJECTION]: The skill processes source code from the convex/ directory which could contain instructions intended to mislead the agent. However, the skill employs deterministic regex-based scanning and follows a predefined 'requireIdentity/requireOwner' pattern, which limits the effectiveness of such injections. Ingestion points: convex/**/*.ts. Capabilities: File modifications and command execution (npx tsc). Sanitization: None explicitly defined for input, but logic is strictly template-based.

convex-authz