Context.dev
Context.dev is a REST API that gives AI agents and software teams web context: any URL returned as clean Markdown, rendered HTML, screenshots or schema-based JSON, plus brand profiles and industry classification from a domain.
What is Context.dev?
Context.dev is a web context API published by Context Dev Inc., a Delaware corporation registered in Wilmington with a team in New York, founded by Yahia Bakour and backed by Y Combinator since May 2026. The stated positioning is the web context API for teams building software and AI agents: send a URL and receive clean Markdown, rendered HTML, a screenshot or structured JSON matching a schema you define; send a domain and receive a typed brand profile, a design system or an industry classification. Four API families sit behind a single key. Web covers scrape, crawl, sitemap, screenshot, parse, search and extract. Brand Intelligence turns a domain, company name, work email or ticker into logos, colours, fonts, postal address, social accounts and firmographic data, and can return a full styleguide. Logo Link is a logo CDN addressed straight from an img tag, without an API call or SDK, on a quota separate from credits. Classification maps a domain to NAICS, SIC and EIC codes or identifies a merchant from a transaction label, while People & News adds news search and person enrichment. Every scrape runs through a real stealth browser: JavaScript rendering, escalation from datacentre to residential proxies, and automatic detection and handling of Cloudflare, DataDome and reCAPTCHA challenges, at no surcharge. Paid plans allow up to five browser actions before capture (wait, scroll, or a natural-language perform instruction) and a country parameter for geolocated fetching through a residential proxy. Markdown output is claimed to be roughly five times more token-efficient than raw HTML, and native documents in PDF, DOCX, DOC, XLSX, PPTX, CSV and XML formats are parsed directly, with OCR billed one credit per recovered page. Access comes through official TypeScript, Python, Ruby, Go and PHP SDKs, an MCP server, a CLI and a skill, plus connectors for Cursor, ChatGPT, Claude, Smithery, Convex, Eve and OpenClaw; Zapier, Make, Google Sheets and Microsoft Excel cover no-code use. Twelve free tools work without an account, and Monitors, announced as a new capability, watch pages for changes. The vendor claims 5K+ developers and names customers including Mintlify, SiteGPT, Sourcely, daily.dev, Similarweb, DocsBot, Bystreet and Adapt. It is explicit about its limits: a five-minute platform timeout, cold requests markedly slower than cached responses with an optional Prefetch, SOC 2 Type I certification with Type II still in progress, and a public status page.
What it does
- Scrape any URL into LLM-ready GitHub Flavored Markdown
- Crawl an entire website with configurable depth, page limits, URL filters and sub-domain following
- Extract structured data against a JSON Schema you define
- Return post-render HTML, page images or a full-page screenshot
- Resolve a domain, company name, work email or ticker into a typed brand profile with logos, colours, fonts, address, social accounts and firmographics
- Parse PDF, DOCX, DOC, XLSX, PPTX, CSV and XML files into clean text
- Monitor pages for changes and search the web or the news with results already scraped
When to use Context.dev / When not to
A quick filter to help you decide if Context.dev is the right fit.
When to use Context.dev
- Product and engineering teams giving an AI agent or an LLM real-time access to the live web
- Teams building RAG pipelines that crawl a sitemap, convert pages to Markdown and push them to embeddings
- Growth, sales and CRM teams enriching records and onboarding forms from a domain or a work email address
- Competitive intelligence and pricing analysts who need page-change monitoring at scale, from 2 to 10,000 concurrent monitors depending on the plan
- Engineering teams retiring in-house scraping infrastructure such as Puppeteer or Playwright workers, proxy pools and CAPTCHA solving
When not to use Context.dev
- Organisations that require an on-premise or self-hosted deployment: the documentation states Context.dev is hosted-only and that there is no self-hosted edition
- Non-technical users with no developer resource: every use goes through an API key and code, or through a Zapier, Make, Google Sheets or Microsoft Excel connector
- Anyone needing a single one-off lookup: the vendor states it does not support one-time lookups, so a subscription is required
- Teams that would have to send sensitive personal data: the Data Processing Addendum instructs customers not to include sensitive information in API requests
- Users who need a localised interface or a mobile client: the site, the documentation and the product are English-only, with no iOS, Android or browser extension
How to use Context.dev
A typical end-to-end flow, from setup to results.
- Create an account on the website and verify your email address; no payment card is required for the Free tier
- Copy the API key issued in the dashboard, in the form ctxt_secret_...
- Alternatively, paste a single line into a coding agent, which signs up, retrieves the key and wires the integration, following the agent authentication guide; a login or email verification is still required
- Install one of the official SDKs for TypeScript, Python, Ruby, Go or PHP, or call the REST endpoints directly on the base URL api.context.dev/v1
- Authenticate every request with the Authorization: Bearer header
- Try each endpoint in the API Playground available in the dashboard before writing code
- Make a first call, for example client.web.webScrapeMd({ url }) in TypeScript, client.web_scrape_md(url=...) in Python, or a GET on /v1/web/scrape/markdown
- For structured extraction, write the JSON Schema that the response must match
- Tune advanced options such as maxAgeMs for cache freshness, includeFrames, settleAnimations, timeoutMS, request tags, Prefetch and batch jobs
- For no-code use, connect through Zapier, Make, Google Sheets or Microsoft Excel; for agents, use the MCP server or a Cursor, ChatGPT or Claude connector
Pros & Cons
Pros
- The stealth stack, covering JavaScript rendering, anti-bot handling and residential proxy escalation, is included at 1 credit per page with no multiplier, against the 5x to 25x multipliers the vendor attributes to competitors
- A single set of APIs and one key cover scraping, structured extraction, brand data and industry classification
- Failed or blocked requests are not billed, and overage is metered in blocks of 10,000 credits and can be switched off rather than causing an abrupt cut-off
- Self-service onboarding: the API key is issued immediately and several named customers report integrating in ten minutes or less
- Official SDKs in five languages, an MCP server, and Zapier, Make, Google Sheets and Microsoft Excel connectors for no-code use
- Standard output formats, including Markdown, your own JSON Schema, NAICS, SIC and EIC codes, PNG and SVG, keep the exit cost low with no proprietary schema
- Documented security posture: SOC 2 Type I, a public DPA with Standard Contractual Clauses, a named sub-processor list and an opt-in Zero Data Retention mode, alongside twelve free tools usable without an account
Cons
- No self-hosted or on-premise edition: the product is hosted-only, so it cannot run inside a customer-controlled perimeter
- The free tier is a one-off allowance of 250 or 500 credits that is never renewed; once it is exhausted the API returns a 401 error
- All purchases are non-refundable under section 7 of the Terms, cancellation only takes effect at the end of the paid period, and there is no per-unit purchase without a subscription
- Cold requests can be markedly slower, up to the five-minute platform timeout, and brand data may be served from a cache up to roughly 90 days old, where a maxAgeMs value of 0 does not force a refresh
- Zero Data Retention is an organisation-level entitlement to be requested rather than a self-service option, and it costs both latency and observability
- SSO/SAML, SCIM, the 99.9% SLA and annual billing are reserved for the Enterprise plan, whose price is not published; support is email-only on Free and Developer, with a Slack channel from Pro upwards
- No Article 27 EU representative is published although the product is sold in Europe, no localised interface is offered beyond English, and the vendor explicitly leaves the legality of each scraping use case to the customer
Pricing & Plans
Context.dev is offered on a freemium basis. A permanent Free plan at 0 USD per month provides a one-off allowance of 250 credits with a personal email address, or 500 with a work address, and requires no payment card. The lowest paid entry point is the Developer plan at 25.00 USD per month, followed by Pro at 149 USD and Scale at 499 USD per month, with Enterprise quoted on request beyond 2 million credits per month. Annual billing grants two months free on every paid plan. Metered overage is charged per block of 10,000 credits at 15 USD on Developer, 9 USD on Pro and 7 USD on Scale. All payments are made in US dollars by Visa, Mastercard, American Express or Discover. One scraped page costs one credit, while advanced endpoints cost 5, 10 or 20 credits.
- 0 USD per month - 250 credits with a personal email address or 500 with a work address
- granted once only
- up to 50 brand retrievals
- 10K Logo Link logos
- 2 concurrent monitors
- 1 concurrent batch
- 10 or 30 calls per minute
- email support
- 25 USD per month - 10
- 000 credits per month
- 15 USD per 10K of overage
- 10
- 000 pages
- 1
- 000 brands
- 1
- 149 USD per month
- presented as the recommended plan - 200
- 000 credits per month
- 9 USD per 10K of overage
- 200
- 000 pages
- 20
- 000 brands
- 499 USD per month - 1
- 000
- 000 credits per month
- 7 USD per 10K of overage
- 1M pages
- 100
- 000 brands
- 100
- on request - more than 2M credits per month
- volume discounts
- custom limits and monitors
- SSO/SAML and SCIM
- a 99.9% SLA
- annual billing
- MSA and DPA
- a dedicated Slack channel and implementation support
Data, GDPR & hosting
A consolidated view of how Context.dev handles your data.
GDPR overview
Context Dev Inc. publishes a Data Processing Addendum incorporating the EU Standard Contractual Clauses (Decision 2021/914) and the UK Addendum. Exhibit A follows Article 28(3) GDPR, Annex II lists security measures and Annex III the sub-processors, with 14 days' notice and a customer right of objection before any sub-processor change. Declared processing locations are the European Economic Area, the United Kingdom and the United States. The DPA commits to processing in accordance with data protection laws, including the GDPR. Data subject rights, namely access, rectification, erasure and withdrawal of consent, are exercised through a DSAR form or privacy@context.dev, alongside the privacy policy effective 20 August 2026. A Comp AI Trust Center reports SOC 2 Type 1 compliance, Type 2 in progress, 25 policies and 47 controls. Two gaps: no Article 27 EU representative and no data protection officer is named. Under-18s are excluded.
Who owns the data?
Under the Terms, the customer keeps ownership of what it submits: Context Dev Inc. asserts no ownership over Contributions and states that users retain full ownership and the associated intellectual property rights. API responses in Markdown, HTML, JSON or images are delivered at request time, and the documentation states there is no vendor-side warehouse to export from in order to leave. For personal data, the Data Processing Addendum makes the customer the controller and Context Dev Inc. the processor. The vendor holds account, organisation, API key, subscription reference and usage history records; card details are handled by the payment provider. Neither the Terms nor the privacy policy effective 20 August 2026 claims ownership of customer output.
Reuse rights
Because the customer retains ownership, output returned by the API can be reused, stored and redistributed without asking Context Dev Inc. for permission; responsibility for the underlying rights, including target sites' terms, copyright, the GDPR and the CCPA, stays with the customer, and the vendor states it gives no legal advice. On its own side, the privacy policy effective 20 August 2026 sets out processing to provide, improve and administer the Services, to communicate, for security and fraud prevention, and for legal compliance. Automatically collected data includes IP address, browser and device characteristics, operating system, language, referrers, country and usage. Trackers fall into three categories, Necessary (always active), Measurement and Marketing, with a consent banner in the EEA, the United Kingdom and Switzerland. Data may be shared with advertising and analytics partners, including a hashed email address for conversion measurement, an activity the policy itself flags as a possible sale or sharing under certain US state laws, with opt-out through Global Privacy Control or the banner. The vendor states it does not process sensitive information, and the DPA prohibits it from selling customer personal information or using it outside the contractual purpose. Model training on customer data is mentioned nowhere on the site.
Data retention & training
Hosting summary
The privacy policy effective 20 August 2026 states that the vendor's servers are located in the United States. Section 7 of Exhibit A to the Data Processing Addendum widens the declared processing locations to the European Economic Area, the United Kingdom and the United States, transfers outside the European Union being covered by the EU Standard Contractual Clauses and the UK Addendum. The sub-processor list published on the site gives each provider's location: Google Cloud Platform and Firebase, Stripe, PostHog, Google Ads and Google Tag Manager, Chatwoot and Resend in the United States, and Plausible Analytics in the European Union, in Ireland. The Trust Center declares eight sub-processors in total, including Cloudflare, Google Cloud and Hetzner. Annex II to the DPA provides for encryption in transit over public networks, encryption at rest being supplied by the underlying cloud providers. In practice, the production perimeter is United States-based; the Irish presence relates to the analytics provider used on the marketing pages, not to customer workloads.
Things to keep in mind
Risks and trade-offs to weigh before adopting Context.dev.
- Free credits are a one-off allowance: once they are spent the API returns a 401 error, and any workflow depending on it stops without warning
- Paid plans renew automatically every month until cancelled, and all purchases are non-refundable, so an unused subscription cannot be recovered
- Brand data may be served from a cache up to roughly 90 days old, and setting maxAgeMs to 0 does not force a refresh: treating cached output as live intelligence leads to decisions taken on stale facts
- Automated extraction returns confident-looking JSON that no human has reviewed; piping it straight into a RAG index or a CRM propagates silent errors at scale and erodes the habit of manually verifying a source
- Cold requests can run to the five-minute platform timeout, and Zero Data Retention must be enabled by the vendor at organisation level beforehand, failing which requests return a 403 ZDR_NOT_ENABLED error
- The privacy policy acknowledges that advertising and analytics sharing may qualify as a sale or a sharing under certain US state laws, and the servers are located in the United States, so transfers out of the European Union must be documented on the customer side
- Legal responsibility for scraping stays with the customer: the vendor points to target sites' terms, copyright, the GDPR and the CCPA and states it gives no legal advice, while the media URLs it returns are external resources the client application must handle defensively
Setup & Integrations
Technical difficulty
Low for anyone comfortable with an HTTP API. Sign-up is self-service and the API key is issued as soon as the email address is verified; a single Authorization: Bearer header is enough. Official SDKs for TypeScript, Python, Ruby, Go and PHP reduce a first call to about three lines, a coding agent can wire the integration from one pasted line, and Zapier, Make, Google Sheets and Excel offer a code-free route. Customers report integrations in two to thirty minutes. The prerequisites are calling an HTTP API or using a connector, and writing a JSON Schema for structured extraction.
Deployment
Integrations
Behind Context.dev
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Context.dev.
Frequently asked questions
Is there a free plan?
How do credits work?
Is there a surcharge for JavaScript rendering, anti-bot bypass or premium proxies?
Which SDKs and integrations are available?
Can a whole website be crawled?
Can I buy a single lookup without subscribing?
How fresh is the brand data?
Is there a mode without data retention?
Where is the data hosted, and is there a self-hosted version?
Is there a minimum age?
Should you pick Context.dev?
Context.dev makes a single, legible bet: consolidate scraping, crawling, structured extraction and brand enrichment behind one API key, and include the expensive part, namely JavaScript rendering, anti-bot handling and residential proxies, in the base price of one credit per page instead of billing it as a multiplier. For a team that would otherwise maintain headless browser workers, proxy pools and CAPTCHA solving, that is a measurable proposition, and the reported ten-minute integrations are consistent with the self-service onboarding. The compliance posture is more mature than the company's age suggests: SOC 2 Type I, a public Data Processing Addendum incorporating the EU Standard Contractual Clauses and the UK Addendum, a named and dated sub-processor list, an opt-in Zero Data Retention mode and a public status page. It remains a young company all the same, backed by Y Combinator since May 2026 with a wall of case studies dated 2025 and 2026, and several assurances are still pending or reserved: SOC 2 Type II is not finished, while the 99.9% SLA, SSO/SAML and SCIM belong to an Enterprise plan whose price is not published. Four reservations deserve weighing. There is no self-hosted edition, so the service cannot run inside a controlled perimeter. Purchases are non-refundable and subscriptions renew automatically. The free tier is a one-off allowance rather than a renewing quota, which makes it a proof of concept and not a production floor. And no Article 27 EU representative is published even though the product is sold in Europe, while the servers themselves sit in the United States. Read together, Context.dev reads as a well-documented infrastructure component for developers who already know they need web data, rather than a turnkey product for occasional or non-technical users.
- Choosing a selection results in a full page refresh.
- Opens in a new window.