Local free
Run Scout on your machine with CLI, HTTP, Docker, or skill usage. You bring compute, browser, storage, and optional keys.
Private beta
CLI / HTTP / Docker / Skill / Hosted beta
Evidence-grade acquisition
Turn messy web pages into citable, downstream-ready records.
Scout is a local-first acquisition workbench for AI workflows. It does more than fetch markdown: it preserves source evidence, blocked pages, citations, validation, and typed records your agents, search indexes, and intelligence pipelines can trust.
Scout beta demo
URL to evidence to records in under a minute.
Watch the beta-safe flow: Scout starts from a public category URL, preserves the evidence trail, produces product records, and exports them to downstream tools.
URL
Evidence
Records
Exports
No hard-site bypass guarantee.
What Scout produces
A run is not done until the evidence is reusable.
Source registry
Every fetched, rendered, saved, blocked, or API-provided source gets a stable identity, URL, provider, status, timestamp, and hash.
Cited records
Records carry citations back to source evidence. Missing citations become validation warnings instead of hidden risk.
Portable artifacts
Outputs are files you can inspect, version, export, hand to agents, load into search, or replay without re-crawling.
Why not just another crawler?
Hosted crawlers return content. Scout returns an evidence trail.
Firecrawl-style APIs are excellent when you need hosted markdown from a URL. Scout is for the next step: turning acquisition into records that can be audited, exported, and reused.
Use Scout when the question is not only "can I fetch this page?" but "can I prove where this field came from, explain what was blocked, and ship the result into a downstream workflow?"
Acquisition ladder
Choose the least expensive mode that works.
Direct / Crawl4AI
Default local acquisition for normal pages, sitemaps, crawls, and screenshots.
Scout browser
Launch a controlled browser, capture screenshots, DOM, links, and source evidence.
User browser
When your browser can see the page, capture authorized session evidence locally.
Saved evidence
Replay saved HTML, DOM, markdown, and artifacts without refetching the website.
Artifact contract
Every run leaves a trail.
run/
manifest.json
records.json
records.jsonl
source_pages.json
blocked_pages.json
validation.json
extraction_report.md
raw/
screenshots/Typed outputs
Records, not blobs.
Exports beyond Algolia
Product data should not be trapped in one destination.
Search-ready
Prepare object IDs, names, prices, categories, URLs, images, variants, citations, and completeness warnings for search indexes.
Spreadsheet-ready
Export product and intelligence records to CSV or Google Sheets workflows for review, enrichment, and lightweight operations.
Database-ready
Keep local SQLite/JSONL datasets for repeatable research, diffing, replay, and agent consumption.
Distribution model
Local first. Hosted optional.
Local Scout
Free for beta users. Run on your machine, choose a workdir, keep artifacts locally, and bring your own optional keys.
- CLI, HTTP API, Docker, and skill usage
- No hosted dependency for normal runs
- Best for private workflows and browser capture
Hosted Scout
Convenience API for teams that want managed workers and API keys. Hosted usage must be metered.
- API key generation and usage credits
- Rate limits, retention limits, and tenant isolation
- Beta pass before full subscription plans
Launch status
Private beta can move. Public launch still has a security gate.
Public launch is blocked
Crawl4AI currently resolves lxml 5.4.0, and the dependency audit reports a known vulnerability fixed in lxml 6.1.0. Scout will not claim public production readiness until the dependency audit is clean, or a formal exception is approved.
Private beta is limited
Hosted beta access is capped, approved-tester only, and metered. Pricing is not approved yet; hosted access is limited credits, not unlimited hosted scraping or lifetime API usage.
Local install remains the primary path
Scout is safest to test locally: your machine, your workdir, your artifacts, your optional keys. No clean security claim is made for hosted production until the remaining launch gates close.
What people use it for
Acquisition primitives for real workflows.
Product catalogs
Extract listing records, preserve source evidence, and prepare exports for Algolia, JSONL, CSV, SQLite, or spreadsheet adapters.
Competitive intelligence
Collect company, investor, careers, news, docs, and website-quality evidence into repeatable artifact folders.
Agent tools
Give local agents a controlled acquisition layer with provenance instead of raw, untraceable fetches.
Pricing direction
Free local. Metered hosted.
Local beta
Install Scout locally and run with your own compute, browser, storage, and optional provider keys.
Hosted beta
Invite-only hosted credits with usage caps and short artifact retention while compute, browser, LLM, storage, security, and support costs are measured.
Pay-as-you-go candidate
Prepaid credits are the leading paid model. Monthly plans come later only if real usage patterns justify them.
Quickstart
Try Scout locally first.
Local Scout is the cleanest beta path because you own the machine, browser, artifacts, and keys. Hosted Scout comes after tenancy, quotas, and payment are production-ready.
git clone https://github.com/arijitchowdhury80/scout.git
cd scout
git checkout codex/scout-platform-foundation
python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
playwright install chromium
scout servePrivate beta
Bring one website you currently research or scrape by hand.
The first beta should test whether Scout makes acquisition more inspectable, reusable, and trustworthy. Not whether it can promise infinite hosted crawling.
Local Scout stays free. Hosted beta creates a metered convenience API key after Stripe confirms payment.