wigolo is in public beta — everything works, and it's improving fast. Hit a rough edge? Tell us

public beta · free · open source

The web, wired
into your local agent.

wigolo is a local-first server that hands any AI agent the whole web — search, fetch, crawl, extract, cache, and research. In your editor over MCP, in your framework through an SDK, or in your self-hosted stack over REST. No API keys. No cloud. No metered bill.

claude code · cursor · codex · gemini cli · vs code · windsurf · zed · antigravity · langchain · crewai · n8n — and any other agent, through MCP, REST, or the SDKs

Get started
your agent — wigolo mcp · no API keys
$0
per query, forever
0
API keys to get started
10
web tools over one MCP
How it works

Everything heavy runs on your machine

No metering, no round-trips you didn't ask for. Your agent reaches the open web through one local process — and the work that costs money elsewhere happens right here instead.

you runyour agent
local processwigolo
reachesthe web

Spoken over MCP — and the same tools answer over a REST API and language SDKs, so anything that talks HTTP can drive it.

fetch

A tiered router that starts at plain HTTP and only escalates on what it actually sees — anti-bot challenges, empty SPA shells — never on a guess about the domain.

1plain HTTP — where most pages resolve
2TLS-impersonation — past bot walls
3a headless browser engine — full render
search

18 search-engine adapters fan out in parallel, then rank fusion and an on-device ML reranker settle the order — with an explainable score behind every single result.

fan out — 18 engines, one query array
rank fusion — merge, dedup, reconcile
on-device rerank — scored, explained
cache

Everything fetched lands in a local index — keyword search paired with on-device vectors. Re-asking is instant and free, and it still answers when you're offline.

keyword + on-device vectors, side by side
a repeat ask returns in milliseconds
works with no network at all
Models and cache live under ~/.wigolo on your machine — no keys, nothing metered, and nothing leaves unless you opt into an LLM.

one command wires it into

+ any other agent, through MCP

The tools

Your agent is smart. Its web access isn't.

Full tool reference
// every result explains itself
{
  "excerpt": "Logical replication replicates data…",
  "score": 0.87,
  "breakdown": {
    "semantic": 0.91,
    "lexical": 0.74,
    "engine_consensus": "4/5 engines"
  },
  "source_span": { "start": 1284, "end": 1571 },
  "citation_id": "f38b44c100e4"
}

Evidence, not blue links

See the receipts

One MCP call fans a query array across many engines in parallel. Every result comes back with a transparent score breakdown, byte-offset source spans, and citation IDs — output your agent can quote.

1
Plain HTTPfast path — most pages end here

↓ anti-bot challenge detected

2
TLS-impersonationlearned fingerprints past bot walls

↓ SPA shell · thin content

3
Headless browserfull render from a warm in-process pool

Routing on observable signals

How it works

The keyless fetch ladder escalates to a real browser on what it sees — SPA markers, challenge bodies, thin content — not domain guesses. It clears most challenge sites with no keys; the strictest managed-challenge and IP-reputation walls need an opt-in proxy, solver, or reader. It learns per-domain, and unlearns when a site stops needing it.

agent → cache "postgres logical replication"
wigolo → 3 hits · hybrid keyword + semantic
   postgresql.org/docs/current/logical-replication
   postgresql.org/docs/current/warm-standby
   wiki.postgresql.org/wiki/Streaming_Replication
latency → 3ms · $0 · nothing left your machine

Everything it sees, it keeps

Get started

Every response lands in a local store under ~/.wigolo/ — full text, keyword index, and on-device vectors. Asking again is instant and costs nothing; find_similar and change detection build on it.

question → "how does postgres logical replication work?"
decompose → 4 sub-queries · fanned out in parallel
    publication / subscription model
   conflict handling & limitations
    vs. physical / streaming replication
synthesize → cited brief · 7 sources · 12 citations
gaps → surfaced, not hidden

Research that writes itself

All 10 tools

research decomposes a question, fans out sub-queries, fetches sources, and hands back a cited brief. agent runs an autonomous plan → search → fetch → extract → synthesize loop with a step log and time budget.

Sponsors

Thank you

wigolo is free for everyone and stays that way. The sponsors below help keep it maintained, and their support goes straight into the work.

  • TestMu AI (formerly LambdaTest) is the world's first full-stack agentic AI quality engineering platform, trusted by 18,000+ enterprises.

    Visit TestMu AI

Want to support wigolo? Sponsorship keeps a free-forever project maintained, and there’s room for more companies and individuals alongside the sponsors above. If that sounds like you, write to ktowhid20@gmail.com — it goes straight to the developer who wrote the code.

Verbatim quoted excerpts, pinned to byte-offset source spans with citation IDs.

All four tools converged on the same core answer. wigolo · the only tool of four to return this

1 / 4
Parity

Same fight.
Different physics.

The paid tools in this lane are genuinely good — that's what makes the parity interesting. What still separates wigolo isn't quality. It's where the work happens.

wigolofirecrawlexatavily
the fight — everyone shows up
Multi-engine web search
Fetch & structured extraction
Whole-site crawl & map
the physics — where the work happens
Verbatim excerpts pinned to byte-offset source spans
Explainable per-result score decomposition
Persistent local memory — instant, offline re-query
Query data stays on your machine
API key / accountnonerequiredrequiredrequired
Cost per query$0meteredmeteredmetered

Feature standing as of July 2026 — check each vendor's docs for current state. One cold query, four tools, judged on the evidence alone — watch the full run.

The meter: a metered cloud API's cost climbs with every query while wigolo stays flat at zero dollars — illustrative pricing
Ask twice, pay never: first query hits the live web in 3.6 seconds, the second answers from local cache in milliseconds, the third still answers offline

Give your agent the whole web.

One instant command wires the local engine into your agent — search, fetch, crawl, extract, cache, and research, with no API key.

1

Install & wire your agent

Auto-wires the agent for you — comma-separated: claude-code · cursor · codex · gemini-cli · opencode · vscode · windsurf · zed · antigravity. Using any other MCP client? Drop --agents — init still sets everything up and you point it at wigolo yourself. init does the full setup — downloads the browser engine + models and runs a health check so any problem surfaces here; add --no-warmup to defer downloads to first use.

That's the whole setup — search, fetch, crawl, extract & cache need no API key. Beyond MCP: a REST API, TypeScript & Python SDKs, an agent-skills installer, and framework integrations (LangChain, CrewAI, LlamaIndex, Vercel AI SDK).

2

Check it's healthy

Verifies the local engine — search, browser, on-device models.

Node ≥ 20 · macOS / Linux / Windows · no account, no key.

3

Optional — answer synthesis

research, agent, and answer-format search use an LLM. Easiest is a free Gemini key (no local setup). Set the provider plus its key — a provider alone isn't enough.

Grab a free key at aistudio.google.com. Prefer fully local & keyless? Run Ollama: export WIGOLO_LLM_PROVIDER=ollama WIGOLO_LLM_MODEL=llama3.1