Skip to content
Toolradar
Kit by Speakeasy logo

Kit by Speakeasy

Claim this toolEditor reviewed

Coding agent runtime that reduces model round trips with a single compose tool

AI CodingTesting & QAAutomation
Visit Website
Tracked since2026
0 reviews tracked

The Bottom Line

Entry price

Free plan available, paid tiers above

Biggest pro

Significantly reduces input token usage and active time compared to other agent harnesses

Biggest con

No built-in security boundary, permissions framework, or sandbox—must be run inside a trusted environment

TL;DR - Kit by Speakeasy

  • Reduces model round trips by exposing a single compose tool that can chain multiple operations (read, edit, test, delegate) in one request
  • Supports open protocols (ACP, A2A, MCP, Agent Skills) and can orchestrate subagents across different harnesses like Claude Code or Codex
  • Small, statically linked binary with long session persistence, automatic context compaction, and crash recovery
Pricing: Free plan available
Best for: Growing teams

What is Kit by Speakeasy?

Editorial review
Kit is a lightweight, open-source coding agent runtime designed to minimize model round trips and token usage. It provides a single tool called "compose" that allows agents to read files, run tests, apply edits, delegate tasks to subagents, and return structured data in one round trip, significantly reducing context repetition and accelerating task completion. Kit includes a terminal client, an Agent Client Protocol (ACP) server, an A2A endpoint, and a subagent orchestrator in a single static binary, enabling seamless integration with compatible editors and other agent harnesses. The runtime supports long sessions with automatic context compaction at 80% of the context window, crash-safe locks, and transcript persistence to disk. It works with various AI models through OpenAI subscriptions, OpenRouter, or the Speakeasy AI Control Plane, and allows changing models mid-session. Kit is designed to run inside a trusted security boundary—it does not include a permissions framework, sandbox, or web UI. It is installable via a script, Docker, or built from source, and is released under an MIT license on GitHub.
how we evaluateSourcegithub.com ↗

Pros & Cons

Pros

  • Significantly reduces input token usage and active time compared to other agent harnesses
  • Flexible model choice and easy switching during a session
  • Open source (MIT) and lightweight with multiple installation options (script, Docker, source)

Cons

  • No built-in security boundary, permissions framework, or sandbox—must be run inside a trusted environment
  • No web UI or graphical interface; only terminal-based interaction

Key Features

Single compose tool for chaining shell, edit, subagent, prompt, fork, and other operations in one round tripReusable subagents with schema validation and support for multiple harnesses over ACPOpen protocol support: ACP v1/v2, A2A v1, MCP, Agent Skills, and Agent Plugin packagesLong session management with automatic context compaction at 80% of context window and crash-safe lock filesModel selection through native OpenAI OAuth, OpenRouter, or Speakeasy AI Control Plane, changeable mid-session

Pricing Plans

Free Trial

Pricing checked Sep 7, 2026

Free

$0 USD per month

  • Unlimited public/private repositories
  • Dependabot security and version updates
  • 2,000 CI/CD minutes/month
  • 500MB of Packages storage
  • Issues & Projects
  • Community support

Team

$4 USD per user/month

  • Everything in Free
  • Access to GitHub Codespaces
  • Repository rules
  • Multiple reviewers in pull requests
  • Draft pull requests
  • Code owners
  • Required reviewers
  • Pages and Wikis

Enterprise

Starting at $21 USD per user/month

  • Everything in Team
  • Data residency
  • Enterprise Managed Users
  • User provisioning through SCIM
  • Enterprise Account to centrally manage multiple organizations
  • Environment protection rules
  • Repository rules (advanced)
  • Audit Log API
Calculate your costView full pricing

Is Kit by Speakeasy worth the price?

84/100

Kit's pricing is generous at the low end: the $0 Free tier includes 2,000 CI/CD minutes and unlimited private repositories, and the $4 per user Team tier is very cheap for managed Git hosting with Codespaces.

The $21 Enterprise starting price is a big jump for audit logs, SCIM, and FedRAMP, but still competitive for security-heavy buyers. This is best for small engineering teams and startups that want modern developer workflow features without a big budget.

Hidden Costs & Gotchas

Team tier caps CI/CD minutes at 3,000 per month, with no published overage rate.

Packages storage is limited to 500MB on Free and 2GB on Team.

GitHub Secret Protection and Code Security cost extra on Team, not included in $4 per user.

Audit Log API and data residency require Enterprise at $21 per user/month.

Free tier lacks Codespaces, required reviewers, and environment secrets, forcing upgrades.

How Kit by Speakeasy's pricing compares

At $4/mo, Kit by Speakeasy is the most affordable of its 2 direct competitors.

Kit by Speakeasy
$4
FlowiseAI
$35
LangChain
$39

Entry paid plan, monthly. Pricing checked Sep 7, 2026.

Reviews

Improve Your Thinking Patterns Using ChatGPT cover
$99Free with your review

Review Kit by Speakeasy, get a free AI guide

Share your experience and we will send you Improve Your Thinking Patterns Using ChatGPT, free.

Write a review

Best Kit by Speakeasy Alternatives

Top alternatives based on features, pricing, and user needs.

n8n logo
n8nFreemium

Workflow automation for technical teams

4.7
LangChain logo
LangChainFreemium

Build LLM-powered applications

4.6
FlowiseAI logo
FlowiseAIFreemium

Visually build, deploy, and scale AI agents and chatbots with an open-source, low-code platform.

4.8
See all AI coding tools →

Still deciding?

Most buyers shortlist 2 or 3 tools before committing. Pull a side-by-side comparison or browse the full alternatives shortlist below.

Kit by Speakeasy vs n8nHead-to-head: features, pricing, who winsKit by Speakeasy vs LangChainHead-to-head: features, pricing, who winsKit by Speakeasy vs FlowiseAIHead-to-head: features, pricing, who wins

Explore More

Best AI Coding ToolsBest Testing & QA ToolsBest Automation ToolsBest Free AI CodingBest Free Testing & QABest Free Automation

Kit by Speakeasy FAQ

What is the compose tool in Kit?

The compose tool is a single function that the model receives to perform all agent actions. A compose program can call shell, edit, subagent, prompt, fork, tool_search, tool, auth, skill, a2a, and docs operations in one round trip, with independent calls running concurrently and data dependencies controlling execution order.

How does Kit reduce token usage compared to other coding agents?

By exposing only one tool (compose) instead of many, Kit avoids multiple model round trips per action. The model can write a short program that reads files, runs tests, applies edits, retries failures, and delegates work—all in a single request, which repeats less context and completes more work per call.

What protocols does Kit support for integration with editors and other tools?

Kit supports Agent Client Protocol (ACP) v1 and v2 over stdio, HTTP/SSE, and WebSocket; A2A v1 in both directions; MCP (Model Context Protocol); Agent Skills; and Agent Plugin packages.

Can I use Kit with agents like Claude Code, Codex, or Cursor?

Yes, Kit supports using Claude Code, Codex, Cursor, or Kit itself as a subagent harness over ACP. You can also spawn subagents that return structured JSON matching a schema.

How does Kit handle long-running sessions and context limits?

Kit synchronizes each append-only JSONL transcript item to disk before accepting it, and automatically compacts context when it reaches 80% of the context window. Sessions can be resumed from the terminal client, prompt, or any ACP client, and Kit will retry eligible subscription failures for up to 24 hours using crash-safe locks.

Is Kit a secure runtime suitable for production use?

Kit is designed as a runtime, not a security boundary. It has no permissions framework, sandbox, or web UI. It should be run inside a trusted security boundary (e.g., a container or VM) and the --root option sets the working directory.

What AI models are supported by Kit?

Kit supports ChatGPT subscriptions through native OAuth, as well as models available through OpenRouter and the Speakeasy AI Control Plane. You can change the model during a session using the /model command.

How do I install Kit and what are the requirements?

Kit can be installed via a curl script (macOS arm64, Linux x86-64), using mise, or via Docker images for linux/amd64 and linux/arm64 with slim, bookworm, and alpine variants. It can also be built from source using the Rust toolchain. The Docker images run as a non-root user and do not include Git or language toolchains by default.

Source: github.com

Guides & Articles

Best Vibe Coding Tools in 2026

Expert guide

Best Open-Source LLMs in 2026

Expert guide

Best AI Coding Tools in 2026

Expert guide