Skip to main content
Glama
SylphxAI

Lookout

by SylphxAI
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Local

Lookout

Zero-config (no install)

npx -y @sylphx/lookout

Starts the MCP server on stdio. CLI:

npx -y @sylphx/lookout doctor
npx -y @sylphx/lookout search "local first agents"

The web from your machine.

Lookout is a local-first web instrument for agents and apps: search, fetch, extract, cache with citeable excerpts — no required API key, tiny default install, same API on SDK · CLI · MCP.

Primary competitive anchor: wigolo (learn multi-surface + honesty; do not require multi-GB browser/model warmup).

Related MCP server: wigolo

Why Lookout wins for agents

  1. Zero-config web instrument — npx -y @sylphx/lookout (MCP on stdio by default).

  2. No API key required for core search/fetch/extract paths.

  3. Citeable excerpts — agents get evidence-shaped results, not anonymous blurbs.

  4. Local-first cache — faster repeat research, fewer surprise network calls.

  5. CLI + MCP — npx -y @sylphx/lookout doctor|search|fetch|extract.

Generic web scraper

Paid search API

Lookout

Agent-ready MCP

❌

sometimes

✅ stdio MCP

Zero-config

rare

account setup

npx -y

API key

optional

required

not required for core

Evidence shape

DIY

vendor JSON

family envelope

Live: @sylphx/lookout@0.2.1 · bin lookout.

Product docs

Doc

Purpose

docs/POSITIONING.md

Strategic positioning

docs/COMPETITIVE.md

Peer anchors and wedge

docs/EVIDENCE_CONTRACT.md

Evidence = result contract

docs/TOOL_SURFACE.md

Few clear tools policy

docs/PRODUCT_INDEPENDENCE.md

This repo is SSOT

docs/IPPB.md

Independent public product bar

docs/NORTH_STAR.md

Customer-value North Star and evidence metric

docs/PUBLISH.md

npm/git publish status

Agent skill surface

See skills/lookout/SKILL.md.

Why Lookout (vs multi-GB “local web” stacks)

Typical agent web path

Lookout

Paid search API keys

Default public adapters, $0/query

Multi-GB browser + model warmup

Tiny default — HTML search/fetch/extract

Opaque snippets

Cite spans + extract route + warnings

One mega-tool

Clear tools: search · fetch · extract · cache (+ advanced crawl)

Primary competitive anchor: wigolo — learn multi-surface honesty; do not require heavy browser/model installs on the default path.

Evidence is a result contract (spans, routes, warnings) — not a tool named evidence_first.

Install (30 seconds)

npm install -g @sylphx/lookout
# brand bin
lookout doctor
lookout search "model context protocol"

Claude Code:

claude mcp add lookout -- npx -y @sylphx/lookout

Install (dev)

git clone https://github.com/SylphxAI/lookout.git
cd lookout
# Bun 1.4.0 (.bun-version)
bun install --frozen-lockfile
./bin/lookout tools
./bin/lookout extract   # via engine with -- see CLI
# CLI
./bin/lookout doctor
./bin/lookout tools
./bin/lookout search "local-first agents"
./bin/lookout fetch https://example.com
./bin/lookout extract https://example.com
./bin/lookout cache stats
./bin/lookout crawl https://example.com --depth 0 --pages 2

# MCP (stdio)
bun src/mcp.ts
# or: ./bin/lookout mcp

SDK

import { Lookout } from '@sylphx/lookout'

const lookout = Lookout.create()
const search = await lookout.search('model context protocol')
const page = await lookout.fetch('https://example.com')
const extracted = await lookout.extract({ url: 'https://example.com' })

Tools (clear, not merged)

Tool

Job

web_search

Public adapters (DuckDuckGo HTML + Wikipedia + npm registry + HN Algolia), rank fusion, query-term boost, host filters

web_fetch

SSRF-safe HTTP(S) fetch, redirects, size limits, cite spans

web_extract

Title, description, JSON-LD, tables, spans from HTML/URL

web_cache (advanced)

Local disk cache query/stats/clear/prune (LOOKOUT_CACHE_DIR, optional max age)

web_crawl (advanced)

Same-origin, depth-limited crawl (not a full-site crawler)

web_research (advanced)

Search then fetch/extract top pages with citeable excerpts

Environment

Var

Purpose

LOOKOUT_CACHE_DIR

Cache directory (default ~/.cache/lookout)

LOOKOUT_CACHE_MAX_AGE_MS

Optional max age for search/fetch cache hits

LOOKOUT_LIVE=1

Enable optional live network tests

LOOKOUT_USER_AGENT

Optional User-Agent override for fetch

LOOKOUT_FETCH_TIMEOUT_MS

Optional fetch timeout (ms)

LOOKOUT_FETCH_MAX_BYTES

Optional max response body bytes

Safety

  • Private IPv4/IPv6, localhost, link-local metadata hosts denied

  • Only http/https

  • Redirect cap + response size cap

Family

Citra · Iris · Cue · Prism · Spine · Lookout

Company portfolio knowledge (docs only — not a product monorepo): This repository is product SSOT. Sibling agent tools are separate repositories and are not vendored here.

License

MIT

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityActive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • wigolo

    KnockOutEZ
    A
    license
    A
    quality
    A
    maintenance
    Provides local-first web intelligence over MCP with tools for search, fetch, crawl, extract, cache, find-similar, research, and autonomous agent loops, requiring no API keys.
    10
    790 npm
    5,227
    AGPL 3.0

  • scrapyard

    artboarding-hash
    A
    license
    A
    quality
    A
    maintenance
    Local MCP server for web search and page extraction, providing clean markdown from URLs, search results, site mapping, and research endpoints without API keys or accounts.
    5
    Apache 2.0

  • web-docs-mcp

    Aleksandrr
    F
    license
    A
    quality
    B
    maintenance
    A local-first MCP server for web search and documentation fetching, providing free DuckDuckGo search, URL-to-markdown conversion, and library docs from npm, PyPI, crates.io, Go, and GitHub with caching.
    5
    -

Appeared in Searches