Skip to content
Back to search

webrouter

WebRouter is an open source standard that connects your agent to the web

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked yesterday

Check it deeper on your machine

It has changed since its scan on 17 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/Nimbleway/webrouter with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/Nimbleway/webrouter'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/nimbleway--webrouter/badge.svg)](/listings/nimbleway--webrouter)

HTML

<a href="/listings/nimbleway--webrouter"><img src="/api/listings/nimbleway--webrouter/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

pip install "webrouter[all]"     # or: uv add "webrouter[all]"

Python 3.11 or later. Extras per method (`search`, `extract`, `research`, `answer`, `browser`, `mcp`, ...) keep the install small; see the [Quickstart](https://docs.webrouter.net/quickstart).

Keys are read from the environment. Copy [`.env.example`](https://github.com/Nimbleway/webrouter/blob/main/.env.example), which names every variable and where to get it, and fill in the ones you hold.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 275 lines
  1. <p align="center">
  2. <img src="https://raw.githubusercontent.com/Nimbleway/webrouter/main/docs/assets/webrouter-mascot.png" alt="WebRouter" width="360">
  3. </p>
  4. <h1 align="center">WebRouter</h1>
  5. <p align="center"><strong>The open source standard for agent web access.</strong><br>
  6. One typed interface for search, extract, answer, research, live data and browsers, across two dozen providers.</p>
  7. <p align="center">
  8. <a href="https://pypi.org/project/webrouter/"><img src="https://img.shields.io/pypi/v/webrouter?color=202c26&label=pypi" alt="PyPI"></a>
  9. <a href="https://github.com/Nimbleway/webrouter/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-Apache%202.0-202c26" alt="License: Apache 2.0"></a>
  10. <a href="https://pypi.org/project/webrouter/"><img src="https://img.shields.io/badge/python-3.11%20%7C%203.12%20%7C%203.13-202c26" alt="Python 3.11 to 3.13"></a>
  11. <a href="https://github.com/Nimbleway/webrouter/actions/workflows/ci.yml"><img src="https://github.com/Nimbleway/webrouter/actions/workflows/ci.yml/badge.svg?branch=main" alt="CI"></a>
  12. <a href="https://discord.gg/MGztFWSjZh"><img src="https://img.shields.io/badge/Discord-5865F2?logo=discord&logoColor=white" alt="Join WebRouter on Discord"></a>
  13. </p>
  14. <p align="center">
  15. <a href="https://docs.webrouter.net"><strong>Documentation</strong></a> ·
  16. <a href="https://docs.webrouter.net/quickstart">Quickstart</a> ·
  17. <a href="https://docs.webrouter.net/providers">Providers</a> ·
  18. <a href="https://github.com/Nimbleway/webrouter/tree/main/examples">Examples</a> ·
  19. <a href="https://discord.gg/MGztFWSjZh">Discord</a>
  20. </p>
  21. > **Developer preview:** the API may change before 1.0. Share feedback on [Discord](https://discord.gg/MGztFWSjZh).
  22. Every web API for agents has its own SDK, parameter names and response shape.
  23. WebRouter puts one interface in front of all of them, so switching provider is
  24. one argument, a failing provider falls back to the next, and your agent code
  25. never changes.
  26. ```python
  27. from web_router import WebRouter
  28. with WebRouter() as router: # no API key needed to start
  29. results = router.search("what is retrieval augmented generation", max_results=5, timeout=10)
  30. pages = router.extract(results.urls[:3], timeout=30) # markdown, whichever backend serves it
  31. print(results.provider, [doc.url for doc in pages.documents])
  32. ```
  33. ## Why WebRouter
  34. - **One call shape, every provider.** The same request and the same typed
  35. response from Tavily, Exa, Brave, Perplexity, Firecrawl, Nimble and the rest.
  36. Nothing is lost: unmapped fields stay in `extra` and the raw payload in `raw`.
  37. - **Never silently weaker.** An option a provider cannot honour raises
  38. `UnsupportedParameterError` instead of being dropped, so a call never runs
  39. under looser constraints than you asked for.
  40. - **Routing based on published benchmarks.** `auto` tries the providers you hold keys for in the
  41. order committed benchmark results rank them, and falls back on an error or an
  42. empty answer. `response.provider` always says who answered.
  43. - **Cost control.** Every response says what it cost and how long it took. Cap
  44. a call or a whole session in dollars and seconds, try the free backends
  45. first, and keep a log of what you spent ([budgets](#budgets-and-cost-control)).
  46. - **Keyless to start.** Search, extract and a local browser work with no API
  47. key. Every key you add unlocks more providers and fallbacks.
  48. - **Built for agents.** An MCP server, plus ready-made tools for LangChain,
  49. Claude and OpenAI agents.
  50. - **Measure, don't guess.** `benchmark()` and `evaluate()` compare providers on
  51. your own queries and score their accuracy against known answers.
  52. - **Yours to run.** Apache 2.0, runs in your process with your own keys, and
  53. sends each key only to its own provider.
  54. ## Install
  55. ```bash
  56. pip install "webrouter[all]" # or: uv add "webrouter[all]"
  57. ```
  58. Python 3.11 or later. Extras per method (`search`, `extract`, `research`,
  59. `answer`, `browser`, `mcp`, ...) keep the install small; see the
  60. [Quickstart](https://docs.webrouter.net/quickstart).
  61. Keys are read from the environment. Copy [`.env.example`](https://github.com/Nimbleway/webrouter/blob/main/.env.example), which
  62. names every variable and where to get it, and fill in the ones you hold.
  63. ## Methods
  64. | Method | What it does | Keyless |
  65. | --- | --- | --- |
  66. | [`search`](https://docs.webrouter.net/search) | Web results for a query | Yes |
  67. | [`extract`](https://docs.webrouter.net/extract) | Page contents for URLs, in markdown, html or text | Yes |
  68. | [`answer`](https://docs.webrouter.net/answer) | A written answer with its sources, in one call | No |
  69. | [`research`](https://docs.webrouter.net/research) | A multi-step research run, as prose or a schema you give | No |
  70. | [`live_data`](https://docs.webrouter.net/live-data) | A vendor's ready-made scraper for one site, returning records | No |
  71. | [`open_browser`](https://docs.webrouter.net/browser) | A browser session you drive, local or in the cloud | Yes (`local`) |
  72. | [`browse`](https://docs.webrouter.net/browser) | A task handed to a hosted browser agent | No |
  73. | [`call_vendor`](https://docs.webrouter.net/vendors) | A vendor's own API (Google Ads, Datadog, ...) with your credentials | No |
  74. ```python
  75. with WebRouter() as router:
  76. router.answer("who won the 2022 world cup final") # .answer, .citations
  77. router.research("how did nvidia's datacenter mix shift in 2025", output_schema=MySchema)
  78. router.live_data("amazon_products", {"url": product_url}, provider="brightdata")
  79. with router.open_browser() as browser: # a local Chromium, no key
  80. browser.goto("https://example.com")
  81. ```
  82. Every method has an async twin on `AsyncWebRouter`.
  83. ## Providers
  84. | Method | Providers |
  85. | --- | --- |
  86. | Search | Nimble, Tavily, Exa, Brave, Perplexity, Parallel, Linkup, You.com, WebIQ, Keenable, DeepSeek, AWS AgentCore, Valyu, Seltz, Octen, Keirolabs, plus the keyless You.com, Exa and Parallel MCP endpoints |
  87. | Extract | Nimble, Tavily, Firecrawl, Exa, Linkup, Oxylabs, Bright Data, Jina Reader, WebIQ, Keenable, DeepSeek, Browserbase |
  88. | Answer | Perplexity, You.com, Linkup, Parallel, Exa, Brave |
  89. | Research | Nimble, Parallel, Perplexity, Linkup, You.com |
  90. | Live data | Nimble, Apify, Bright Data, Oxylabs |
  91. | Browser sessions | Local Chromium, Browserbase, Steel, Hyperbrowser, Anchor, Kernel |
  92. | Browser agents | Browserbase, Browser Use, Skyvern |
  93. [Providers](https://docs.webrouter.net/providers) has each provider's key, the
  94. options it maps and how it was verified. Adding one is two methods: see
  95. [Adding a provider](https://docs.webrouter.net/custom-providers).
  96. ## Routing and fallback
  97. For search, extract, answer, research and the browser methods, name no
  98. provider and `auto` serves the call: every provider you hold a key for, in the
  99. order committed [benchmark results](benchmarks/providers/README.md) rank them,
  100. then the free tiers. A member that fails, is rate-limited or answers empty
  101. passes the call to the next. A member that cannot honour an option you set is
  102. skipped for that call, and if none can, the call raises rather than running
  103. without it. Live data has no fallback, since a source belongs to one vendor:
  104. `live_data_sources()` with no provider searches every vendor you hold a key
  105. for, and each source it returns names its own, which `live_data()` runs it on.
  106. Vendor calls always name their vendor, since an account belongs to one.
  107. ```python
  108. from web_router import Provider, WebRouter
  109. WebRouter(Provider.TAVILY).search("...") # one provider, by name
  110. WebRouter().search("...", provider="exa") # or per call
  111. router = WebRouter(provider_config={"auto": {"chain": {"extract": ["nimble", "firecrawl"]}}})
  112. router.extract(url) # Nimble, then Firecrawl if Nimble fails
  113. ```
  114. A named provider is never quietly swapped for another: a missing key raises an
  115. error naming the variable to set. In a service, share one router per process,
  116. set `timeout=` on each call and turn on the `web_router` logger: see
  117. [Running in production](https://docs.webrouter.net/production). [The auto order](https://docs.webrouter.net/providers#the-auto-order)
  118. lists each method's order and how it is generated.
  119. ## Budgets and cost control
  120. ```python
  121. from web_router import Budget, WebRouter, prices
  122. router = WebRouter(router="cheapest_first") # free backends first, then the cheapest paid ones
  123. log = router.log_spend("spend.jsonl") # from here on, every answered call is logged
  124. response = router.answer("...", budget=Budget(max_cost_usd=0.01, time_limit_s=10))
  125. response.provider, response.cost_usd, response.cost_source, response.elapsed_s
  126. with router.budget(Budget(max_cost_usd=3.0, time_limit_s=600)) as b:
  127. b.search("...")
  128. b.research("...")
  129. b.status() # spent, reserved, remaining, time left
  130. log.report() # what you spent, by day, backend and method
  131. prices.estimate("tavily", "search") # 0.008: list price of one request
  132. ```
  133. Every response carries `cost_usd`: the backend's own figure where it states
  134. one (`cost_source="reported"`), else an estimate. Unless a router ranks the
  135. backends itself (the default `rules` router does not), a cost cap makes `auto`
  136. try the cheaper of the backends that score alike first.
  137. Each attempt is checked against what is left before it starts, so parallel
  138. calls never overspend together, and a backend that does not fit is skipped
  139. with the reason recorded. Estimates come from the list prices shipped in
  140. `web_router.prices`, not your bill. Only search (its keyless backends) and
  141. `open_browser` (`local`) have a $0 backend; [Cost and budgets](https://docs.webrouter.net/budgets)
  142. has the table per method, a fall-back for a batch that runs out, and the spend log. An
  143. experimental router can also pick the method and provider for each prompt, and
  144. `url_policy="public_only"` keeps a planned call off private and cloud-metadata
  145. hosts. See [Smart routing](https://docs.webrouter.net/smart-routing).
  146. ## Use it from an agent
  147. **MCP server.** Every method as a tool, running on your machine with your keys:
  148. ```bash
  149. claude mcp add web-router -- uvx --from "webrouter[mcp]" web-router-mcp
  150. ```
  151. The [MCP guide](https://docs.webrouter.net/mcp-server) covers Claude Desktop, Cursor and other clients,
  152. credentials, tool costs and browser guards.
  153. **Agent frameworks.** Ready-made `web_search` and `web_extract` tools, with a
  154. compact rendering for the model and the full response kept for you:
  155. ```python
  156. from web_router import WebRouter
  157. from web_router.integrations.langchain import web_router_tools
  158. tools = web_router_tools(WebRouter(), max_results=5) # LangChain
  159. ```
  160. `web_router.integrations.anthropic` and `web_router.integrations.openai` give
  161. Claude and OpenAI agents the same tools as plain tool definitions. See
  162. [Agent integrations](https://docs.webrouter.net/integrations).
  163. ## Compare providers on your own queries
  164. ```python
  165. report = router.benchmark("nvidia q3 earnings", providers=["nimble", "tavily", "you-mcp"])
  166. for run in report.runs:
  167. print(run.provider, run.latency, run.count if run.ok else run.error)
  168. ```
  169. `benchmark_extract()` does the same for extract, and `evaluate()` scores each
  170. provider's accuracy on a question set with known answers (SimpleQA, FRAMES,
  171. SealQA or your own CSV) with an LLM judge. See
  172. [Benchmarks and evaluation](https://docs.webrouter.net/benchmarks), and
  173. [`benchmarks/`](benchmarks/) for the published runs behind the default order.
  174. ## WebRouter or calling each provider directly
  175. | | Each provider's SDK | WebRouter |
  176. | --- | --- | --- |
  177. | Switching provider | Rewrite the call, options and parsing | One argument |
  178. | Response shape | Different per provider | One typed model, raw payload kept |
  179. | An unsupported option | Varies by provider | Raises, naming who can serve it |
  180. | A provider fails | Your retry code | Falls back to the next you hold a key for |
  181. | Which provider to use | Guesswork | Order from published benchmarks, `benchmark()` on your queries |
  182. | Spend | Per-vendor dashboards | Per-call and per-session caps, list prices in the package |
  183. | Agent tools | One integration per provider | One MCP server and one set of tools |
  184. ## Documentation
  185. | Get started | Methods | Go further |
  186. | --- | --- | --- |
  187. | [Introduction](https://docs.webrouter.net) | [Search](https://docs.webrouter.net/search) | [Agent integrations](https://docs.webrouter.net/integrations) |
  188. | [Quickstart](https://docs.webrouter.net/quickstart) | [Extract](https://docs.webrouter.net/extract) | [Smart routing](https://docs.webrouter.net/smart-routing) |
  189. | [How it works](https://docs.webrouter.net/how-it-works) | [Answer](https://docs.webrouter.net/answer) | [Benchmarks](https://docs.webrouter.net/benchmarks) |
  190. | [Providers](https://docs.webrouter.net/providers) | [Research](https://docs.webrouter.net/research) | [Adding a provider](https://docs.webrouter.net/custom-providers) |
  191. | [Configuration](https://docs.webrouter.net/configuration) | [Live data](https://docs.webrouter.net/live-data) | [MCP server](https://docs.webrouter.net/mcp-server) |
  192. | [Errors](https://docs.webrouter.net/errors) | [Browser](https://docs.webrouter.net/browser) | [Examples](https://github.com/Nimbleway/webrouter/tree/main/examples) |
  193. | | [Vendor data](https://docs.webrouter.net/vendors) | [Changelog](https://github.com/Nimbleway/webrouter/blob/main/CHANGELOG.md) |
  194. ## Contributing
  195. Contributions are welcome, and the most useful one is small: map one more of a
  196. provider's options, or add a provider.
  197. ```bash
  198. make install # every extra, dev tools and pre-commit hooks
  199. make check # lint, typecheck and the offline unit tests; no keys needed
  200. ```
  201. [CONTRIBUTING.md](CONTRIBUTING.md) has the workflow and the provider checklist,
  202. [STYLE_GUIDE.md](STYLE_GUIDE.md) the review standards, and
  203. [AGENTS.md](AGENTS.md) the rules for AI coding agents working on this repo.
  204. ## Community
  205. - [Discord](https://discord.gg/MGztFWSjZh): questions in #help, and what you built in #showcase.
  206. - [GitHub Issues](https://github.com/Nimbleway/webrouter/issues): bugs and gaps in a provider's support.
  207. - [GitHub Discussions](https://github.com/Nimbleway/webrouter/discussions): proposals and design ideas.
  208. - [SECURITY.md](SECURITY.md): vulnerabilities. Never in a public channel.
  209. <p align="center"><em>Useful? Star the repo so more agent developers can find it.</em> ⭐</p>
  210. ## License
  211. [Apache License 2.0](LICENSE). The SimpleQA grader prompt vendored at
  212. `src/web_router/evals/prompts/` keeps its own MIT license, in
  213. `LICENSE-simple-evals` beside it.
  214. <!-- mcp-name: io.github.Nimbleway/webrouter -->

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.