scottschindler

Scan a local codebase for Agent Observe evidence and gaps. Use when a user asks to map or audit AI agent prompts, Vercel AI SDK calls, OpenAI SDK calls, tool definitions, tool schemas, chain steps, API routes, client chat entrypoints, eval coverage, trace IDs, logging, loop bounds, needsApproval gates, experimental_telemetry, or side-effect risks in a repository.

Scan a local codebase for Agent Observe evidence and gaps. Use when a user asks to map or audit AI agent prompts, Vercel AI SDK calls, OpenAI SDK calls, tool definitions, tool schemas, chain steps, API routes, client chat entrypoints, eval coverage, trace IDs, logging, loop bounds, needsApproval gates, experimental_telemetry, or side-effect risks in a repository.

Design and build a working agent with Vercel's eve framework. Use when the user wants to build, plan, or architect an AI agent with eve — choosing the right eve features (tools, skills, connections, channels, subagents, schedules, sandbox), then building it step by step with auth, approvals, and evals. Complements the official vercel/eve skill, which covers API reference via bundled docs.

Use this skill when the user has a product idea, startup idea, app idea, feature idea, vague concept, rough notes, or scattered thoughts and wants to turn it into a clear, specific, buildable product spec, MVP plan, PRD, user flow, feature list, launch plan, or Cursor-ready build prompt. This skill should clarify the idea, ask high-leverage product questions when needed, make reasonable assumptions when appropriate, and produce an opinionated product spec. Do not give generic brainstorming when this skill applies.

Answer Tempo blockchain questions using official documentation. Use when asked about Tempo protocol, TIP-20 tokens, fees, transactions, stablecoin DEX, or any Tempo-related questions.