BrowserBash
Free open-source CLI that turns plain-English objectives into real Chrome browser tests with an exit-code verdict.
The pitch is rare because the free tier is the whole product, not a teaser — full CLI, local dashboard, unlimited local runs, no locked commands. Where it earns its keep is the parts a thin wrapper wouldn't bother building: Verify steps that compile to Playwright checks, an MCP server that hands your agent a structured verdict, and `--budget-usd` to stop a runaway suite. The one real catch is cloud retention — free accounts keep uploaded run history for 15 days and then auto-delete, so anything
Verified 5d ago · liveness 81/100 · cite: rightaichoice.com/tools/browserbash
- AI agent developers who need an MCP validation layer returning structured verdicts instead of guesses
- QA engineers replacing brittle CSS selectors and XPath with committable plain-English test files
- SDETs building maintainable E2E suites with deterministic Playwright-backed assertions
- Teams wiring browser validation into CI with exit codes, NDJSON output and per-suite budgets
- Teams that need free cloud run history kept longer than 15 days — extended retention is coming soon
- Users who want a fully managed SaaS with nothing installed — the CLI is self-installed via npm
- Tests requiring fine-grained control over browser events or custom wait conditions
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 BrowserBash if you need a fully managed SaaS you never install, or if you want cloud run history held permanently on a free account — the free tier auto-deletes runs after 15 days and extended retention is still listed as coming soon.
Free cloud run history is capped at 15 days and then auto-deletes (v1.3.1, 12 Jun 2026), so anything you need to keep longer must be archived to your own storage or to --upload destinations.
At $0 forever, BrowserBash undercuts paid AI browser platforms that require a credit balance or their own cloud — the entire CLI, local dashboard, and a cloud account cost nothing, with no credit card and no trial clock. Cheaper than any metered cloud alternative for teams running local Chrome and Ollama.
In short
BrowserBash — Free open-source CLI that turns plain-English objectives into real Chrome browser tests with an exit-code verdict. Best for AI agent developers who need an MCP validation layer returning structured verdicts instead of guesses, QA engineers replacing brittle CSS selectors and XPath with committable plain-English test files, SDETs building maintainable E2E suites with deterministic Playwright-backed assertions. Free to use.
What's new in BrowserBash
Checked 6 days agoAcross the latest 5 updates: 4 feature updates and 1 launch.
v1.3.1 — New local dashboard command, --upload flag, 15-day free cloud retention
Adds a free web dashboard that runs entirely on your machine to browse runs, a --upload flag to push a single run to the cloud on demand, and sets free cloud run retention to 15 days with auto-delete after.
v1.3.0 — Expiring API keys, per-account rate cap, connection health verification
Adds 30-day expiring API keys for linking the CLI to your account so stale credentials roll over automatically, a per-account rate cap to keep cloud usage predictable, and health verification for the CLI-to-cloud connection.
v1.2.0 — Session recording and per-run replay
Adds --record to capture video and screenshots of a run, stores video and screenshot artifacts in the cloud when a run is uploaded, and adds per-run replay in the dashboard so you can step through what the agent did.
v1.1.0 — Run sync to the cloud dashboard
Links the CLI to an account with browserbash connect to send runs to your account, adds a runs API powering the sync flow, and adds a runs view in the cloud dashboard for synced run history.
v1.0.0 — First public release
First public release: plain-English objectives drive a real local Chrome with no code or selectors, markdown *_test.md test files with @import composition, Ollama, OpenRouter and Anthropic backends, NDJSON agent mode with exit codes 0/1/2/3, and a free local web dashboard.
What people actually say about BrowserBash — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
13 mentions across 2 sources (YouTube, Product Hunt) · researched Aug 25, 2026.
Average across the 2 sources that answered — each source counts once, not each post.
- +Zero API keys or credit cards needed — works with Ollama or free OpenRouter models.
- +Plain-English 'say what to test' eliminates brittle CSS selectors and XPath.
- +CLI-first approach keeps testers close to existing shell workflows.
- +MCP server plugs into Claude Code, Cursor, and Codex in one line.
- +Deterministic Verify assertions compile to real Playwright checks — no model in loop.
- −Default engine is Stagehand, so 'different from Stagehand' is a fair question.
- −Questions about intent preservation after UI changes remain unanswered.
- −Monetization model unclear — users wonder how it stays free.
- −Very few real-world user reports beyond the initial Product Hunt launch.
- −No long-term reliability data for CI pipeline gating.
- • No hidden costs yet, but optional paid cloud run retention is planned
- • Using paid models (Anthropic, OpenAI, Google) costs API fees if you opt in
- • Cloud run retention beyond free 15 days will cost money soon
Viability Score
How well maintained and how widely used is BrowserBash? 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: October 2026
How we score →Key Features
- Plain-English objectives drive a real Chrome browser end to end, no selectors or code
- Built-in MCP server (browserbash mcp) exposing run_objective, run_test_file and run_suite over stdio
- Markdown *_test.md tests with @import composition and variable templating (testmd v2, per-step execution)
- Deterministic Verify assertions that compile to real Playwright checks with expected-vs-actual evidence
- API steps for data seeding over plain HTTP inside test files
- Saved logins via auth save / --auth replaying a session across tests
- Session recording with --record capturing video and screenshots (v1.2.0)
- Per-run replay in the cloud dashboard for stepping through a run
- Cloud run sync via browserbash connect plus a runs API (v1.1.0)
- Expiring 30-day API keys and CLI-to-cloud connection health verification (v1.3.0)
- NDJSON --agent mode with a stable, additive one-event-per-line schema
- CI exit codes 0/1/2/3 for pipeline gating
- cost_usd per run and --budget-usd per suite with hard stop and skipped-test reporting
- Replay cache re-runs green tests with zero model calls
- Parallel sharded runs with --shard and viewport matrices
About BrowserBash
BrowserBash is a free, open-source command-line tool that turns one plain-English sentence into a real Chrome browser run and returns a verdict your pipeline can read — a status, extracted values, deterministic assertion results, and cost in one JSON object. You install it with `npm install -g browserbash-cli` and no signup is required to run. The default stack is your local Chrome, the MIT-licensed Stagehand engine, and a local Ollama model, so there are no API keys and no credit card involved. Recent releases shipped a run-sync cloud dashboard with per-run video replay (v1.2.0), expiring API keys and health checks for the CLI-to-cloud connection (v1.3.0), and a local dashboard command (v1.3.1). Every layer is swappable: point the same objective at any CDP endpoint, Browserbase, LambdaTest, or BrowserStack; bring an Anthropic or OpenAI key, or run any OpenAI-compatible server such as vLLM or LM Studio. Tests live in committable `*_test.md` files with `@import` composition and variable templating. Verify steps compile to deterministic Playwright checks with no model in the loop and ship expected-vs-actual evidence on failure, while agent-judged checks are clearly flagged. API steps seed data over plain HTTP inside the same file. A built-in MCP server (`browserbash mcp` on stdio) exposes `run_objective`, `run_test_file`, and `run_suite` to Claude Code, Cursor, Codex, or any MCP host, giving coding agents a validation loop rather than a guess. `--agent` streams one JSON event per line on a schema that has held since v1.0. Budgets are enforced with `cost_usd` per run and `--budget-usd` per suite, the replay cache re-runs green tests with zero model calls, and CI gating uses exit codes 0/1/2/3. It is aimed at QA engineers and SDETs tired of maintaining selectors and page objects, and at AI-agent developers who need to prove a change worked before merging. The closest technical neighbor is Stagehand, which BrowserBash wraps in a CLI, a test format, budgets, and MCP
Behind the Verdict
Most AI browser tools want a credit balance or a driver class before you see anything move. BrowserBash goes the other way: `ollama pull qwen3 && browserbash run "..."` gives you a real Chromium session on a local model with nothing metered. In practice that's the fastest way to evaluate whether plain-English testing suits your team at all, because you find out in a minute without a sales call. Pick this when you're a QA engineer or SDET replacing CSS selectors and page objects, or an agent developer who needs a hook that returns pass/fail plus extracted values instead of a screenshot you have to eyeball. The MCP server is the standout: `claude mcp add browserbash -- browserbash mcp` and your coding agent can ask "did my change break the checkout?" as a tool call. The replay cache matters more than it sounds — agents can validate after every edit without burning tokens. Pass when your test needs fine-grained control over browser events or custom wait conditions. The English layer abstracts those away by design, and when a flow depends on a precise timing trick you'll fight it. Teams without Node installed, or those who want a fully managed SaaS with nothing local, should look at a cloud AI-browser platform instead. The real-world caveat is retention. Free cloud accounts keep run history for 15 days then delete it automatically; the paid Supporter tier that extends that window is listed as coming soon, so today every account gets the same 15 days. Anything you need to keep has to live locally or in your own CI artifacts. Runs also stay on your machine unless you add `--upload`, which is the right default but means cross-team visibility takes deliberate setup. Versus Stagehand — its own default engine — BrowserBash is the opinionated packaging: a test file format,
Researching BrowserBash? 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 BrowserBash actually fits — and what changes day-one when you adopt it.
Install with npm install -g browserbash-cli, run one plain-English objective against local Chrome with Ollama to confirm the loop, then move the flows into committable *_test.md files with @import and Verify steps that compile to Playwright checks.
Outcome: Deterministic expected-vs-actual evidence on failures and a suite that lives in the repo instead of in a page-object layer.
Add the MCP server to Claude Code or Cursor with `claude mcp add browserbash -- browserbash mcp`, then have the agent call run_objective against the checkout flow and read status, summary and assertion results from the returned JSON.
Outcome: A merge gate that returns an exit code and evidence rather than a model's opinion, with the replay cache keeping green paths at zero model calls.
Wire the CLI into GitHub Actions or Jenkins, gate on exit codes 0/1/2/3, split the suite across machines with --shard, and set --budget-usd so an overnight run hard-stops instead of spending without limit.
Outcome: Browser validation in the pipeline with a spend ceiling and skipped tests reported rather than silently dropped.
Use Cases
- Prove an AI coding agent's change didn't break checkout before you merge a PR, via an MCP tool call returning a structured verdict.
- Replace brittle CSS selectors and XPath with committable markdown *_test.md suites that compose shared flows with @import.
- Gate CI pipelines on exit codes 0/1/2/3 in GitHub Actions or Jenkins instead of parsing test output.
- Capture a login once with auth save and replay it with --auth so tests stop hitting re-login walls in CI.
- Record a session with --record to get video and screenshots for debugging or for sharing repro steps.
- Turn an existing Playwright suite into plain-English test twins with browserbash import and review what it generated.
- Run the same test as a production check with browserbash monitor --every 10m, alerting Slack only when pass flips to fail.
- Check responsive layouts across viewports in one command with --matrix-viewport.
Models Under the Hood
as of 2026-10-15
Limitations
- BrowserBash is a free, open-source (Apache-2.0) CLI installed via npm (npm install -g browserbash-cli) that turns plain-English objectives into real Chrome runs.
- It runs on free local models via Ollama or free OpenRouter models with zero API keys; bringing your own Anthropic or OpenRouter key is optional.
- The free cloud account keeps run history for 15 days, after which runs auto-delete (v1.3.1, 12 Jun 2026); extended retention beyond 15 days is listed as coming soon.
- No signup is required to run the CLI — an account is only needed for the cloud dashboard.
as of 2026-10-02
Verification history
We have re-verified BrowserBash 7 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-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 7 verification passes.
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 BrowserBash tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free
$0/forever
Ideal for
Solo QA engineers, SDETs and agent developers who want the whole CLI and cloud account at no cost, running local Chrome with Ollama or free OpenRouter models.
What this tier adds
Starting tier — every command with no locked features, all providers, zero API keys, markdown tests, NDJSON agent mode, exit codes 0/1/2/3, local dashboard, and a cloud account with 15-day run retention.
Supporter
Custom (coming soon)
Ideal for
People and teams who want their uploaded cloud run history kept past the free 15-day window and who want to fund ongoing Apache-2.0 development.
What this tier adds
Adds extended cloud run retention beyond the free 15 days on top of everything in Free — listed as coming soon, with no published price and every account on 15-day retention for now.
Where the pricing makes sense
The company stage and team size where BrowserBash's pricing actually pencils out — and where peers do it cheaper.
At $0 forever, BrowserBash undercuts paid AI browser platforms that require a credit balance or their own cloud — the entire CLI, local dashboard, and a cloud account cost nothing, with no credit card and no trial clock. Cheaper than any metered cloud alternative for teams running local Chrome and Ollama.
Setup time & first value
How long it actually takes to get something useful out of BrowserBash — broken out by persona, not the marketing-page minute.
Fastest path is minutes: `npm install -g browserbash-cli`, `ollama pull qwen3`, then `browserbash run "…"` — no signup, no keys, no credit card, and the site claims you can automate in about sixty seconds. Adding the MCP server to Claude Code is one line. Cloud sync needs a free account via `browserbash connect`. Wiring CI exit codes and sharding takes a bit longer. Swapping to Browserbase,
Switching to or from BrowserBash
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Playwright: run `browserbash import` to convert your Playwright suite into plain-English test twins, then review and edit the generated markdown before cutting over.
- →From Stagehand: the default engine already is Stagehand, so move your objectives into *_test.md files and pick up the CLI, budgets and MCP tools around the same engine.
- →From a paid AI browser platform: keep the same objectives, point the provider at local Chrome and the LLM at Ollama, and run with no credit balance while you compare verdicts.
- →From manual QA scripts: capture the login once with auth save, then write the flow as a committable markdown test with API steps for any data seeding.
- ↗To Playwright: Verify steps already compile to Playwright checks, so rebuild the flow as native tests if you need fine-grained browser-event control.
- ↗To Stagehand directly: drop the CLI wrapper and call the MIT library from your own harness if you want the engine without the test format.
- ↗To a cloud-only platform: re-create objectives on Browserbase, LambdaTest or BrowserStack as the provider if you decide you want managed infrastructure.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “BrowserBash”, and we withheld 6: 6 could not be judged, because “BrowserBash” 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 BrowserBash.
Official links
Tools that pair well with BrowserBash
Common stack mates teams adopt alongside BrowserBash, with the specific reason each pairing earns its keep.
Featured Head-to-Head Comparisons
Browserbash vs Locus Robotics
For cost-free, open-source, natural-language browser automation with zero API keys, pick BrowserBash. For scalable, AI-driven warehouse robotics that boost picking productivity by 2-3x and reduce labor, choose Locus Robotics. They solve completely different problems: digital QA vs. physical fulfillment.
Browserbash vs Truleo
Truleo and BrowserBash serve entirely different domains. If you're in law enforcement needing to connect siloed data for case leads, Truleo’s specialized intelligence platform is worth the investment. For browser automation in QA or CI/CD pipelines, BrowserBash is a free, open-source, zero-API-key solution that just got a local dashboard and 15-day cloud retention. Choose based on your job function, not features.
Browserbash vs Presto Voice
If you're a QA engineer or developer needing free, open-source browser automation that works with local or cloud AI models, BrowserBash is a no-brainer. If you run a QSR chain and want to automate drive-thru ordering with proven revenue lift, Presto Voice is purpose-built for that. These tools serve completely different markets — choose based on your domain.
Alternatives to BrowserBash
View all
Open Interpreter
Open Interpreter is an open-source terminal coding agent that turns plain-English requests into real file edits and shell commands.
Popular in Browser & Computer-Use Agents
Spider Cloud
Spider Cloud is a web scraping, crawling, and search API that streams live pages as markdown or JSON for AI agents and RAG pipelines.
OpenAgents
OpenAgents is an open-source language agent platform for data analysis, 200+ plugins and autonomous web browsing.
Frequently Asked Questions
Used BrowserBash? Help shape our editorial sentiment research.