Kit
AIAn open-source coding agent runtime that gives the model a single compose tool, cutting round trips and token use versus conventional harnesses.
Kit
An open-source coding agent runtime that gives the model a single compose tool, cutting round trips and token use versus conventional harnesses.
About Kit
Kit is a coding agent runtime distributed as a single static binary that bundles a terminal client, an Agent Client Protocol (ACP) server, an A2A endpoint, and a subagent orchestrator. Where most harnesses expose many tools and burn a model round trip on each call, Kit exposes exactly one tool called compose, whose argument is a short program written in the Runlet language. A single compose program can read files, run tests, apply edits, retry failed commands, delegate to subagents, and return structured data in one round trip, which is why Speakeasy measures roughly half the input tokens and active time per hand-written line compared with Codex CLI or Claude Code on size-matched production tasks. Because it speaks ACP v1 and v2 over stdio, HTTP/SSE, and WebSocket, Kit works inside compatible editors and can orchestrate other harnesses — including Claude Code, Codex, and Cursor — as subagents without custom integrations. It also supports A2A v1 in both directions, MCP servers, Agent Skills, and Agent Plugin packages, and it is built for long sessions: every transcript item is synced to an append-only JSONL file before it is accepted, context compacts automatically at 80% of the window, and crash-safe locks let a session resume from the TUI or any ACP client.
Key Features
Use Cases
Frequently asked questions about Kit
What is Kit?
An open-source coding agent runtime that gives the model a single compose tool, cutting round trips and token use versus conventional harnesses.
How does Kit work?
Kit works by combining Single Compose Tool: The model receives one tool whose argument is a Runlet program, so file reads, shell commands, edits, retries, and subagent calls all happen in a single round trip instead of one per action., Concurrent Program Execution: Independent calls inside a compose program run concurrently, with data dependencies or after blocks used to force ordering when needed., Reusable Subagents: A subagent is a value you can continue, fork, inspect, and close, and you can require its output to match a JSON schema., Cross-Harness Orchestration: Claude Code, Codex, Cursor, or another Kit instance can be used as the subagent harness over ACP, so Kit coordinates tools it did not ship with., Open Protocol Support: Implements ACP v1 and v2 over stdio, HTTP/SSE, and WebSocket, A2A v1 in both directions, plus MCP, Agent Skills, and Agent Plugin packages. to help users with Cost-Sensitive Agentic Coding: Cut token spend and wall-clock time on large refactors by collapsing many tool calls into one composed program., Editor-Integrated Agents: Drive Kit from any ACP-compatible editor without writing a bespoke integration for each client., Multi-Harness Pipelines: Orchestrate Claude Code, Codex, or Cursor as subagents from a single controlling program when different harnesses suit different steps., Long-Running Migrations: Run multi-hour codebase migrations that survive crashes and resume from a durable transcript., Structured Extraction from Code: Require subagents to return schema-validated JSON so results can be fed into downstream tooling rather than parsed from prose..
What are the main features of Kit?
Key features include Single Compose Tool: The model receives one tool whose argument is a Runlet program, so file reads, shell commands, edits, retries, and subagent calls all happen in a single round trip instead of one per action., Concurrent Program Execution: Independent calls inside a compose program run concurrently, with data dependencies or after blocks used to force ordering when needed., Reusable Subagents: A subagent is a value you can continue, fork, inspect, and close, and you can require its output to match a JSON schema., Cross-Harness Orchestration: Claude Code, Codex, Cursor, or another Kit instance can be used as the subagent harness over ACP, so Kit coordinates tools it did not ship with., Open Protocol Support: Implements ACP v1 and v2 over stdio, HTTP/SSE, and WebSocket, A2A v1 in both directions, plus MCP, Agent Skills, and Agent Plugin packages..
Who is Kit for?
Kit is useful for anyone interested in Cost-Sensitive Agentic Coding: Cut token spend and wall-clock time on large refactors by collapsing many tool calls into one composed program., Editor-Integrated Agents: Drive Kit from any ACP-compatible editor without writing a bespoke integration for each client., Multi-Harness Pipelines: Orchestrate Claude Code, Codex, or Cursor as subagents from a single controlling program when different harnesses suit different steps., Long-Running Migrations: Run multi-hour codebase migrations that survive crashes and resume from a durable transcript., Structured Extraction from Code: Require subagents to return schema-validated JSON so results can be fed into downstream tooling rather than parsed from prose..
How much does Kit cost?
Kit is free to use.
How do I get started with Kit?
Visit https://github.com/speakeasy-api/kit to sign up and explore Kit.
Explore more AI Ai Agents tools
Browse all Ai Agents tools →Browse by use case: Code Generation · Automation & Productivity
Compare Kit: vs Tadata · vs Reflexio · vs HyperProbe · vs OpenMAIC