The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run shell commands using npx, specifically npx tsc --noEmit and npx convex dev --once. These are standard development workflows for the Convex platform intended to verify code compilation and deployment readiness.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest local project data to inform its behavior, which presents a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: Reads project-specific files including convex/schema.ts and convex/_generated/ai/guidelines.md.
Capability inventory: The agent can execute shell commands via npx and read/write files within the convex/ directory.
Boundary markers: No specific delimiters are used to separate ingested file content from the system instructions.
Sanitization: The skill does not implement specific sanitization or validation of the ingested file content.
[EXTERNAL_DOWNLOADS]: The skill references and encourages the use of official vendor packages for implementing specific functionalities.
Evidence: Directs the agent to use @convex-dev/agent and @convex-dev/workflow for chat features and multi-step flows.

convex-expert