agentskit.js
Open specs

Open specs

Three portable JSON contracts — Agent-to-Agent, Skill+Tool Manifest, Eval Format.

Small, stable, versioned JSON shapes. Each is a subpath of @agentskit/core: types + validator + zero runtime dep.

  • A2A (Agent-to-Agent) — agent/card · task/invoke · task/cancel · task/approve · task/status. JSON-RPC 2.0. Recipe.
  • Manifest — packaging format for skills + tools. Tool entries mirror MCP inputSchema so manifests round-trip.
  • Eval Format — portable eval dataset + run-result. matchesExpectation supports literal / regex / normalized / semantic similarity.

Per-spec deep dives land in step 6 of the docs IA rollout.

✎ Edit this page on GitHub·Found a problem? Open an issue →·How to contribute →
Edit on GitHub

CLI

Nine commands wrapping every part of AgentsKit — init, chat, run, dev, doctor, ai, tunnel, rag, config.

Recipes

Copy-paste solutions for common scenarios. Each recipe is end-to-end and runs as written.

On this page

Related