Keystroke
Build internal AI agents and multi-step systems that run as real TypeScript in your own project.
Keystroke is a strong fit if you want deployable internal agents rather than a chat wrapper. It writes real TypeScript (deployed from your own repo on Hobby), gives agents memory, a filesystem with bash, and 1,000+ integrations, and only bills when agents and workflows run ($0.01/run, Exa search at $0.007). Three build surfaces — the platform agent, your coding agent, or MCP — mean coders and non-coders share one project. Choose Windmill or n8n if you want a purely visual, non-code automation tool, and Zapier if you only need simple app-to-app steps. Choose Keystroke if your workflows need reasoning agents with sandboxed code and durable replay.
Verified 15d ago · liveness 71/100 · cite: rightaichoice.com/tools/keystroke
- Engineering teams that want agents as reviewable TypeScript
- Ops, sales, and finance teams with a technical partner nearby
- Startups shipping internal automation on metered pricing
- Companies already using Cursor or Claude Code and wanting a deploy target
- Teams that want a drag-and-drop, no-code-only builder
- Solo users who just need a simple chatbot
- Organizations that require SSO and audit logs on a sub-$100 plan
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Keystroke if you want a purely drag-and-drop automation tool and your team has no appetite for repos, TypeScript, and git — or if you need SSO, audit logs, and role-based access on a plan below Organization.
Everything is metered after your credit: agent and workflow runs cost $0.01 each, so a workflow polling every few minutes adds up fast
Hobby's $1/mo usage credit makes it a genuine exploration tier; Pro at $20/mo with $20 included credit suits a small team shipping real internal agents. Both are cheaper than a seat-priced enterprise agent platform, but a heavy always-on workflow can outrun the credit quickly. Organization is quote-based and is where SSO, audit logs, RBAC, guardrails, and unlimited concurrency live.
In short
Keystroke — Build internal AI agents and multi-step systems that run as real TypeScript in your own project. Best for Engineering teams that want agents as reviewable TypeScript, Ops, sales, and finance teams with a technical partner nearby, Startups shipping internal automation on metered pricing. Free to start; paid plans from $20/mo.
What's new in Keystroke
Checked todayAcross the latest 1 update: 1 feature update.
What people actually say about Keystroke — is it worth it?
We scanned public community sources for Keystroke on Aug 6, 2026 and could not establish that the discussion we found is about this tool rather than something else sharing its name. Our own analysis of that scan says the posts were off-subject. Rather than publish a sentiment score built on the wrong subject, we publish nothing here and re-run the scan.
Viability Score
How well maintained and how widely used is Keystroke? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: September 2026
How we score →Key Features
- Platform agent that builds, tests, and deploys agents from chat with no IDE required
- Open TypeScript framework and headless CLI for Cursor, Claude Code, and Codex
- MCP connection to build from ChatGPT, Claude, or any MCP client
- Agent memory and persistent workspace
- Built-in filesystem with bash and sandbox compute
- Database access for agents
- Web search and web fetch via Exa
- Browser use for agents
- Custom tools and skills
- Durable workflows with step replay and typed Zod input/output
- Durable sleeps and human approval hooks
- Triggers: schedules, webhooks, and events from connected apps
- 1,000+ integrations plus any HTTP API and MCP servers
- External channels including Slack, Microsoft Teams, Telegram, Discord, Linear, WhatsApp
- Instant apps for agent-built dashboards and internal tools
About Keystroke
Keystroke is an all-in-one platform for building internal AI agents and AI systems. You describe what you want, and a built-in platform agent writes it as ordinary TypeScript in your project, then runs it with integrations, credentials, run history, and a shared team workspace. There are three ways to build: the in-app platform agent (no IDE, CLI, or repo needed), your own coding agent such as Cursor or Claude Code against Keystroke's open TypeScript framework and headless CLI, or an MCP connection so you can build from ChatGPT or Claude chat. Agents get persistent memory, a filesystem with bash, a database, web search, browser use, and instant apps for lightweight dashboards. Systems combine agents, durable workflows, triggers, and human approval steps, with run traces that show what happened across agents, tools, and human decisions. Keystroke covers 300+ models through an AI gateway, 1,000+ integrations (Slack, Google, Linear and hundreds more), any HTTP API, and MCP servers. It fits technical and non-technical teams that want agents doing real work inside their existing tools rather than a standalone chatbot.
Behind the Verdict
Keystroke's pitch is that everything you build is real code in a real project, and that is its main differentiator. Whether you chat with the platform agent, drive it from Cursor or Claude Code via the headless CLI, or connect over MCP from ChatGPT or Claude, you get the same TypeScript output that the other surfaces can read, test, review, and maintain. That matters because agent platforms that store logic in a proprietary visual graph are hard to review, version, and hand off. On capabilities, the agent runtime is unusually complete for a tool in this price band: persistent memory, a built-in workspace with files and bash, custom tools, skills, project files, self-scheduling triggers, web search and web fetch through Exa, browser use, and external channels including Slack. Systems layer adds durable step replay, action/agent/LLM steps, typed input and output with Zod, durable sleeps and approval hooks, and trigger-driven runs. The run-trace view spans agents, workflows, tools, and human decisions, which is what you need before letting an agent touch production. Pricing is metered and transparent, which is rarer than it should be. Hobby is free with 5GB storage, 5 concurrent runs, and 60 requests/min. Pro is $20/mo and includes $20 of usage credit, 50GB storage, 10 concurrent runs, 300 requests/min, 30-day log retention, 30-minute agent and workflow timeouts, and 5 projects. Runs cost $0.01 each, empty poll triggers $0.005, Exa search $0.007, Exa fetch $0.001, and sandbox compute about $0.067/hour, all charged at cost. LLM calls go through an AI gateway at a 1.1x markup on the provider's listed price, or you bring your own keys and pay nothing extra. The Hobby tier's $1/mo credit is a real ceiling, not a trial. Where it does not fit: if you want a no-code visual builder with drag-and-drop nodes, Keystroke expects you to be comfortable with repos, git, TypeScript, and tests — or at least to lean on the platform agent. If you are a solo user wanting a simple chatbot, this is more machinery than you need. Governance features teams will ask about first — SSO and identity controls, audit logs, role-based access controls, agent guardrails, unlimited concurrency, and enterprise-grade security — sit behind the Organization plan, and Company Brain is listed as coming soon. There are no published SLA or uptime figures in the material reviewed.
Researching Keystroke? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Keystroke actually fits — and what changes day-one when you adopt it.
Connect the repo to Keystroke, describe an agent that summarizes merged pull requests, let the platform agent write it in TypeScript, run the evals, then deploy with the CLI on a webhook trigger from the main branch.
Outcome: A deployed agent that posts shipped-work summaries automatically, with run history in the shared workspace and a pull-request diff the team can review like any other code.
Build an agent that reads call notes when they are ready, checks whether pricing was agreed, posts to Slack, creates a Linear issue, and pauses for human approval before sending the order form.
Outcome: Call follow-up runs reliably with a run trace showing the Slack message, the Linear issue, the page update, and who approved the next step.
Create an agent named Timmy with Superhuman tools wired for listing and searching threads, drafting replies, and managing labels, plus a Slack gateway for requests on the go.
Outcome: Inbox triage and batch drafting run on demand from Slack, with drafts created by default and sending only when explicitly asked.
Use Cases
- Triage an inbox and draft replies in your voice using an agent wired to Superhuman mail tools
- Turn a merged pull request into a shipped-work summary via a trigger on the main branch
- Confirm whether pricing was agreed on a customer call and post the answer to Slack and Linear
- Automate a support desk with an agent you run evals against before production
- Spin up an instant internal dashboard for pipeline, revenue, and team performance
- Build a morning briefing that pulls from connected apps on a schedule
- Give every team one place to share agents, skills, and company context
Models Under the Hood
as of 2026-09-14
Limitations
- Everything you build is real TypeScript in your own project, so you need comfort with repos, git, and tests — or you lean on the platform agent/MCP to write, test, and deploy it.
- Governance features are gated on the Organization plan: role-based access controls, agent guardrails, advanced usage & spend limits, unlimited concurrency, and enterprise-grade security.
- Hobby's included usage credit is $1/mo (Pro is $20/mo + additional usage with $20 of included credit), with runs and polls charged per use and Exa search/fetch billed at cost.
- Company Brain is listed as coming soon with no published release date.
as of 2026-09-14
Verification history
We have re-verified Keystroke 3 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-checked, vendor evidence unchanged
- — re-checked, vendor evidence unchanged
- — re-checked, vendor evidence unchanged
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Keystroke tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Hobby
$0/mo
Ideal for
A developer or small team exploring Keystroke with one project, a repo of their own, and light experimentation
What this tier adds
Free entry point: 1 project, 5GB storage, 5 concurrent runs, 60 requests/min, 7-day log retention, and a $1/mo usage credit
Pro
$20/mo
Ideal for
A small team shipping internal agents and workflows into production and needing more headroom and longer traces
What this tier adds
Adds $20 of included usage credit, 5 projects, 50GB storage, 10 concurrent runs, 300 requests/min, 30-day log retention, and 30-minute agent and workflow timeouts
Organization
Custom
Ideal for
Companies with security and compliance requirements that need centralized access control and support
What this tier adds
Adds advanced usage and spend limits, role-based access controls, agent guardrails, unlimited concurrency, enterprise-grade security, and priority support
Where the pricing makes sense
The company stage and team size where Keystroke's pricing actually pencils out — and where peers do it cheaper.
Hobby's $1/mo usage credit makes it a genuine exploration tier; Pro at $20/mo with $20 included credit suits a small team shipping real internal agents. Both are cheaper than a seat-priced enterprise agent platform, but a heavy always-on workflow can outrun the credit quickly. Organization is quote-based and is where SSO, audit logs, RBAC, guardrails, and unlimited concurrency live.
Setup time & first value
How long it actually takes to get something useful out of Keystroke — broken out by persona, not the marketing-page minute.
Hobby with the platform agent: minutes to a first built agent, since no IDE or repo is required. Building from your own coding agent: under an hour if you already have Cursor or Claude Code and a git repo, since the CLI and framework are TypeScript you can test normally. MCP from ChatGPT or Claude: minutes in a hosted workspace. Wiring integrations, credentials, and evals before production: a day
Switching to or from Keystroke
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From n8n or Zapier: rebuild logic as TypeScript agents and workflows using the platform agent, then keep app connections through the 1,000+ integrations
- →From a custom prompt script: run it through the AI gateway or bring your own keys, and move scheduling to Keystroke triggers
- →From a boutique agent vendor: connect to any HTTP API or private internal API and manage OAuth and API keys in one place
- ↗To self-hosted orchestration: the agents and workflows are ordinary TypeScript in your project, so you can take the code with you
- ↗To a no-code tool: expect to rewrite logic as visual nodes, since Keystroke's durable replay and approval hooks are code-level constructs
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Keystroke”, and we withheld 6: 6 could not be judged, because “Keystroke” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Keystroke.
Official links
Tools that pair well with Keystroke
Common stack mates teams adopt alongside Keystroke, with the specific reason each pairing earns its keep.
Notable
Notable runs healthcare AI agents that automate patient access, revenue cycle, care operations, and contact center work for health systems.
Cargo
Code-first GTM infrastructure for AI agents — define plays, tools, and data models in TypeScript, git, and CLI.
AutoGen Studio
Open-source, low-code GUI for prototyping multi-agent AI systems on Microsoft's AutoGen.
Featured Head-to-Head Comparisons
Keystroke vs Temporal Ai
If you need to rapidly build AI workflows with a visual editor and connect to common SaaS tools, Keystroke is your pick. But if your agents must survive crashes, retries, and long-running processes, Temporal's durable execution is what you truly need for production reliability.
Keystroke vs Air Ai
If you're a developer building custom AI workflows, Keystroke is the agile, cost-effective choice—it's freemium and integrates with the AI models you already use. But if you're in defense or government and need to compress materiel release timelines and achieve mission-critical readiness, Air AI is the specialized powerhouse—recently winning a $31M contract and expanding its Pittsburgh office, it's clearly the defense-first platform. Pick based on your domain, not just features.
Keystroke vs Cryptohopper
If you're a developer building custom AI agents and want flexibility to orchestrate models and APIs, Keystroke is your platform. If you're a crypto trader or a developer who needs live market data feed to AI agents, Cryptohopper is the clear choice — its MCP server and recent guides show a strong focus on AI-integrated trading. They're not direct competitors; pick based on your domain.
Alternatives to Keystroke
View all
Notable
Notable runs healthcare AI agents that automate patient access, revenue cycle, care operations, and contact center work for health systems.
Cargo
Code-first GTM infrastructure for AI agents — define plays, tools, and data models in TypeScript, git, and CLI.
AutoGen Studio
Open-source, low-code GUI for prototyping multi-agent AI systems on Microsoft's AutoGen.
Frequently Asked Questions
Used Keystroke? Help shape our editorial sentiment research.