Free · Source available · Windows · macOS · Linux
Opens in a third of a second. Free to use, source available. No account needed. No additional AI required.
Performance
No sign-in, no cloud sync, no waiting. Every number here was measured, and the benchmark is a script you can run yourself.
Requests per second, running 500 requests that each have a pre-request script and two tests. Higher is better.
| Runner | Total time | Requests/s | Peak memory |
|---|---|---|---|
| HttpDash 1.4.0 | 0.32 s | 1,545 | 107 MB |
| Postman's Newman 6.2.3 | 5.14 s | 97 | 253 MB |
| Bruno CLI 4.2.1 | 43.78 s | 11 | 427 MB |
Each runner ran the same 500 GET requests against a local JSON endpoint. Every request had a pre-request script that sets a variable, and a test script with two assertions. The figures are the median of five runs after a warm-up, with peak memory from /usr/bin/time. Measured in October 2026 on an Apple M5 Mac with 16 GB of memory. Start-up time is from launch to a window showing the workspace.
To reproduce it, clone the repository and run npm run bench. See the benchmark script.
Built for real APIs
Point a collection at its login request once. Before every call, HttpDash checks the token: if it's missing or about to expire, it runs the login first, saves the new token, and sends your request. If a server still answers 401, it refreshes and retries.
data.jwTokenexpires_in
Connect Consul, Eureka, Kubernetes or your own JSON endpoint, and HttpDash creates an environment with one variable per service, like {{orders_url}}. Re-sync whenever services move: addresses update, new services appear, and the variables you added yourself stay put.
Send to each instance of a service in turn and HttpDash counts what every one of them answers: requests, p50 and p95 times, errors. Give a request several payloads and it counts those too, so a pod that is slow for everyone can be told apart from an order that is slow on every pod.
X-Served-By
Give a request, a folder or a whole collection a health URL, and HttpDash reads it before the request goes out, while it is in flight and after it answers. The Health tab shows what moved while your request ran: connections held, queues, a database pool filling up, a service that stopped saying it was healthy.
/healthz is read on the instance the request went to
One click sends the same request to every instance behind a service and puts the answers side by side. HttpDash finds the instances in your registry, or in every address the host name resolves to, and points out the one that answered differently: an old version, a missing field, a different state.
Every response shows where it went on the network: the name's aliases and every address it resolves to, the address it connected to split into octets with its segment, its reverse-DNS name, and the adapter your request left from. One click traces the route, router by router.
Services on a plant network, in a data center or inside a cluster are one tunnel away, without a VPN. Point HttpDash at an SSH jump host or a SOCKS5 proxy, list the hosts it carries, and every request to them goes through it, with their names resolved on the far side.
~/.ssh/config, and never waits for a password*.mes.corp, 10.42.0.0/16) or put a whole environment through one tunnel
Claude and other AI agents can use HttpDash the way you do: find requests, send them, add and update them, start runs and read the results, and look at the screen. They connect through an OpenAPI document or MCP, on your own computer only, and you decide what they are allowed to do.
httpdash mcp for MCP clients
Each request you send by hand is kept as a version, with the response it got. Switch back to one in a click, rename it, tag it and find it in search. When the OpenAPI document the request came from changes, HttpDash tells you what no longer fits before you send it.
Give requests as many tags as you like, and HttpDash suggests the ones you already use. The search box knows every name, URL, method and tag in your workspace, so you pick from a list instead of remembering what you called it.
#smoke and “pay”httpdash run backup.json --tag smoke
Requests leave from a native HTTP client, not a browser, so every header goes out as typed, including Referer, Origin, User-Agent, Host and Cookie. The timeline shows the exact request that was sent and how long each step took.
Switch a response to Table and arrays of objects become sortable tables, with a column for every key. Nested objects and arrays open in place, so you can drill into items or address without leaving the row.
Connect to ws:// and wss:// servers, send text or JSON, and watch everything that goes each way with timings and sizes. The handshake carries your headers and the collection's auth, so authenticated sockets just work.
{{variables}} apply
When an API streams Server-Sent Events, as AI APIs do, HttpDash shows each event the moment it arrives instead of waiting for the end. Stop whenever you like and keep what came in.
Everything in one app
Folders, tabs, history, globals and secret variables with {{autocomplete}} and hover previews. Give each environment a color so you always know where you are sending.
Pre-request and test scripts with a familiar pm API, Chai assertions, JSON Schema checks and a snippet library.
Pick the requests to run from a tree or by tag, then iterations, CSV/JSON data files, a gap in milliseconds between requests and a detailed, exportable report.
Serve saved examples on a local port, pick responses with a header, and switch an environment to the mock in one click.
Run a collection every few minutes and get a desktop notification when it fails, and again when it recovers.
Turn any request into cURL, fetch, Axios, Python, Go, PHP, C#, Java, PowerShell, wget or raw HTTP.
Bring in Postman collections, OpenAPI and Swagger (JSON or YAML) and cURL. Export everything as a backup, or as portable collection files.
Share collections and environments through any git repository. One readable file per request, merges that just work, and secrets that never leave your computer.
Connect, send text or JSON, and follow a live log of every message, with your headers and auth in the handshake.
Server-Sent Events appear as they arrive, so streaming and AI APIs are easy to watch and debug.
Read JSON arrays as sortable tables and export them to Excel, CSV or an image.
A collection that follows your OpenAPI document. Each new version shows what changed, operation by operation, before you apply it.
Send to the first instance, each in turn or a random one, from Consul, Eureka or Kubernetes pods. Unhealthy instances are skipped.
Give a request the variations it should be called with, typed in or imported from CSV or JSON, and send them in turn or at random.
Requests, timings and errors per instance and per payload, live, with slow and failing ones flagged. Export to Excel, PDF or an image.
Send one request to every instance of a service, from your registry or DNS, and see which one answered differently.
An OpenAPI document and an MCP server let AI agents search, send, edit and run requests, with permissions you set and a history of every call.
Every request you send by hand is kept with its response. Rename, tag and search versions, and get a warning when the spec no longer fits.
Any number of tags per request, suggested as you type, and one search across names, URLs, methods and tags.
Read a service's health endpoint before, during and after each request, and see which fields moved while it answered.
Every address a name resolves to, the one it went to in octets and segments, your adapter, and the route hop by hop.
Run the login first when the token is missing or expired, save the new token and send your request, without copying anything.
Environments built from Consul, Eureka, Kubernetes or an OpenAPI document, and kept in step with every sync.
Reach private networks through an SSH jump host or a SOCKS5 proxy, by host or for a whole environment, without a VPN.
Run collections in CI with JUnit reports, tags, data files and tunnels, or send one request like curl with httpdash send.
Finds keys, tokens and passwords typed straight into requests and moves them into masked variables in one click.
Require a directory (LDAP) sign-in to open HttpDash, limit it to a group, and lock it when idle.
Mutual TLS per host, your company's CA, and certificates already installed on your computer, trusted as in a browser.
Automation
Command line
Export your workspace without secrets, commit it, and run it anywhere Node.js runs. Pass credentials as variables, get JUnit reports your pipeline already understands, and fail the build when a test fails.
httpdash send: one request like curl, through the running app, kept with --save$ npm install -g github:gethttpdash/httpdash
$ httpdash run backup.json -e "QA" --env-var password=$PASS -r cli,junit
HttpDash Orders API (environment: QA)
→ List orders
GET https://qa.example.com/orders [200 OK, 8.6KB, 91ms]
↳ ran "Get access token" first (expired) 200
✓ Status is 200
✓ Returns a list of orders
Total time 412 ms. PASSEDSecurity & privacy
HttpDash has no account and no cloud sync. Collections, environments and history live in a folder on your computer.
Require an LDAP sign-in to open the app, limit it to certain users or a group, and lock it after a period of inactivity. While locked, the app shows no data at all.
The secret scan spots keys, tokens and passwords typed straight into requests and moves them into masked variables with one click.
Exports can leave out every secret, token and password, while keeping {{variable}} references intact.
A local audit log notes sign-ins, exports, imports, runs and requests, keeping only method, host, path and status.
Download
Pick your platform. Your collections, environments and settings carry over when you update.
Windows 10 and 11, x64 and ARM64. Installer or portable app.
Download for WindowsNot signed yet: if SmartScreen appears, choose More info → Run anyway.
Apple Silicon and Intel. Disk image or zip.
Download for macOSNot notarized yet: if macOS blocks it, open System Settings → Privacy & Security and click Open Anyway.
x64 and ARM64. AppImage for any distribution, or a .deb for Debian and Ubuntu.
AppImage: chmod +x HttpDash-*.AppImage, then run it.
For CI servers and schedulers. Needs Node.js 22 or later.
npm install -g github:gethttpdash/httpdashThen run httpdash run --help.
Free, and staying that way
No account, no ads, no paid tier. Buy me a coffee if HttpDash saves you time, or sponsor it as a company.