# Firecrawl for Small Business: Web Data, Real Costs, and Where Permission Stops

What Firecrawl does, its hosted prices and credit math, five small business uses, a permission checklist, and the CFAA, privacy, contract, and AGPL risks.

Author: J.A. Watte
Published: November 3, 2026
Source: https://jwatte.com/blog/firecrawl-web-data-api-small-business/

---

**Firecrawl turns web pages into text or organized records that software and AI assistants can use. It suits a small business with a specific, permitted collection job, such as preparing its own help pages for a support assistant. The biggest risk is treating a successful download as permission to collect or reuse the content. Try a small, capped job on your own public pages first; settle site permissions, Firecrawl's commercial terms, and data handling before expanding.**

This expands the Firecrawl entry in our [GitHub tools guide](/blog/github-top-repos-2026-small-business/). We read the live README, selected source code, licenses, documentation, policies, and pricing on November 3, 2026. We did not install or run Firecrawl. The examples below are illustrations, not customer stories or measured performance results.

## What it does, in ordinary words

An **API**, short for application programming interface, is a defined way for one program to ask another for work. Think of a restaurant order slip: give Firecrawl a page address and an output format, and receive the result. An **API key** is the private credential that identifies your account and lets requests spend its allowance.

**Markdown** is ordinary text with simple marks for headings, links, and lists. Turning a page into markdown means extracting readable content from its HTML, the code that describes a web page, while preserving useful structure. An AI assistant can then read a warranty heading and the paragraphs below it without also reading every menu and styling instruction. Cleaning the page does not establish that the warranty is current or that the assistant understood it. Keep the source address and observation time beside the text.

Firecrawl sells a **hosted service**, meaning its computers do the work for you. It also publishes code for **self hosting**, where your IT person operates the service on a computer you control. These routes have different features and responsibilities. Its [README](https://github.com/firecrawl/firecrawl#readme) advertises handling JavaScript, the code that makes pages interactive, and rotating network addresses through intermediary servers, including handling blocked content. Those are vendor claims about capability. We recommend against using them to get past a site that has said no.

An **endpoint** is a particular API operation, like a separate order form for each service:

| Operation | What you ask for | Useful distinction |
|---|---|---|
| [Scrape](https://docs.firecrawl.dev/features/scrape) | Read one specified page | Start here when you know the address |
| [Map](https://docs.firecrawl.dev/features/map) | Discover page addresses on a site | Produces a list to review, rather than all page contents |
| [Crawl](https://docs.firecrawl.dev/features/crawl) | Follow links and collect pages within a scope | Can multiply the job; set an explicit page limit |
| [Search](https://docs.firecrawl.dev/features/search) | Find relevant results across the web | Review results before collecting their full pages |
| [Agent](https://docs.firecrawl.dev/features/agent) | Describe a data gathering task in words | The service decides where to look; harder to constrain and budget |
| [Change tracking](https://docs.firecrawl.dev/features/change-tracking) | Compare a page with an earlier snapshot | A difference needs a person to interpret it |

The README describes Agent as the evolution of its Extract endpoint. We use Agent for that comparison here, rather than presenting older extraction examples as the current starting point. **JSON** means named fields in a machine readable record, such as a price and currency. Extracting those fields with AI can cost more than obtaining the page's markdown.

[Monitor](https://docs.firecrawl.dev/features/monitoring) adds scheduled checks and notifications. It differs from asking for change tracking during individual Scrape calls. A **webhook** is a notification sent to another program when something happens. Receiving a change notification should create a review task, not automatically rewrite your prices or announce that a competitor changed its policy.

## Hosted prices and the credit arithmetic

A **credit** is Firecrawl's usage unit. It is not a dollar, and one request can spend several credits. The following is the full public plan lineup in US dollars before tax from the [pricing page](https://www.firecrawl.dev/pricing), read at 21:34 UTC and rendered in a browser at 21:36 UTC on November 3, 2026.

| Plan | Monthly billing | Annual billing, displayed monthly equivalent | Credits each month | Concurrent browsers | Extra credits per $5 |
|---|---:|---:|---:|---:|---:|
| Free | $0 | $0 | 1,000 | 2 | Not available |
| Hobby | $19 | $16 | 5,000 | 5 | 1,000 |
| Standard | $99 | $83 | 100,000 | 25 | 2,000 |
| Growth | $399 | $333 | 500,000 | 50 | 2,500 |
| Scale | $749 | $599 | 1,000,000 | 100 | 5,000 |
| Enterprise | Custom | Custom | Custom | Custom | Custom |

**Concurrency** is how many browser page renders can happen at once, separate from the number of API requests allowed each minute. It is capacity, not permission to send that much traffic to somebody's website. Annual figures above are the site's displayed monthly equivalents, billed yearly. The displayed annual savings and rounded monthly figures do not reproduce an exact annual invoice consistently. Confirm the full annual charge before buying; we did not enter checkout.

The same page gives these credit rules, supplemented by the [billing documentation](https://docs.firecrawl.dev/billing):

| Work | Published credit rule |
|---|---|
| Basic Scrape or Crawl | 1 per page |
| Map | 1 per call in its dedicated docs and pricing FAQ |
| Search | 2 per 10 results; collecting result pages adds their scrape costs |
| JSON, Question, or Highlight formats on Scrape/Crawl | 4 additional credits per page per selected format |
| [Interact](https://docs.firecrawl.dev/features/interact), the browser interaction feature | 2 per browser minute for code only, 7 with an AI prompt; one minute minimum; zero-data-retention option adds 2; initial scrape billed separately |
| Monitor | 1 per page per check; its docs add 1 for each changed page evaluated by its AI judge |
| Monitor deterministic extraction | Pricing summary lists a flat 7 per page per check |
| Agent | 5 free daily runs advertised; further usage has dynamic pricing |

There is a documentation mismatch worth keeping visible: the pricing page's small endpoint table labels Map as per page, while its summary, FAQ, and [Map documentation](https://docs.firecrawl.dev/features/map) say per call. Our worksheet uses per call and asks the buyer to verify it. We do not promise a fixed dollar price for an Agent task. A returned error page can still consume a basic credit; the pricing FAQ distinguishes that from a scrape returning no result.

Pay-as-you-go adds credits in $5 purchases on paid plans. The FAQ says a monthly limit of `0` switches it off, a blank limit leaves it unlimited, and manually purchased credits do not count toward that limit. Check the setting yourself. Unused ordinary credits do not roll over on Hobby, Standard, or Growth. The pricing FAQ gives Scale one month of rollover and Enterprise custom rollover, while billing documentation adds annual-plan conditions; confirm eligibility for your billing term. Purchased extra credits expire on subscription cancellation. These details matter more to a small pilot than maximum throughput.

## Five small business jobs with permission and budgets

Every quantity here is an assumed workload for an illustration. These estimates cover Firecrawl only. They exclude staff time, any separate AI assistant subscription, storage, and workflow software. No paid AI plan is required merely to call the web data API; the cost of the assistant you connect is a separate purchase we have not priced here.

### Your own help pages for a support assistant

For example, a repair shop could collect 500 of its own public service and warranty pages once a month, then give reviewed text to its support assistant. It could also prepare an `llms.txt` file, a plain text guide pointing AI tools toward useful site material. Markdown output alone does not create that guide or guarantee that any assistant will use it.

Permission basis: the business owns the pages and has rights to the included material. Exclude customer portals, booking details, and licensed content that cannot be reused. Map once, approve the addresses, then scrape markdown with a 500 page ceiling. `1 Map call × 1 + 500 pages × 1 = 501 credits`, within Free's 1,000 monthly allowance if nothing else spends it. Have the warranty owner approve the result before the assistant answers customers. [Map](https://docs.firecrawl.dev/features/map), [Scrape](https://docs.firecrawl.dev/features/scrape), [prices](https://www.firecrawl.dev/pricing).

### Public vendor or competitor changes

For example, a bicycle retailer could check 40 public price and policy pages four times in a budgeting month. Permission basis: the target permits this use in its terms or grants written permission. Being a competitor does not create that permission. Prefer a vendor's feed or licensed API when available.

Use only the approved addresses, retain source dates, and stop on any denial. With plain markdown and basic change tracking, `40 × 4 × 1 = 160 credits`, within Free. The [change-tracking docs](https://docs.firecrawl.dev/features/change-tracking) say basic and text-difference tracking do not add to the scrape charge; JSON tracking costs 5 credits per page. They also say snapshots persist without expiring. That retention deserves approval before using customer or personal information. A person checks whether a changed number is a price, tax, promotion, or layout artifact before acting.

### RFP and grant opportunities

For example, a small building contractor could check 25 agency listing pages four times in the month for requests for proposals, or RFPs, and grants. Permission basis: each agency allows the collection and intended reuse. A public agency address alone does not settle terms or rights in attached documents.

`25 × 4 × 1 = 100 credits`, within Free for basic page collection. Keep the title, agency, closing date, and official link. Exclude login systems, applicant records, and unnecessary contact details. PDF attachments and additional pages need their own scope and credit budget. A staff member checks the actual notice and amendments before bidding. [Scrape](https://docs.firecrawl.dev/features/scrape), [prices](https://www.firecrawl.dev/pricing).

### Research for content you write

For example, an accounting firm could run 20 searches returning 10 results each, then approve and read 100 selected public pages. Permission basis: each selected source allows the collection; any quotation or reuse also needs a copyright basis. Finding a page through search does not grant a republication license.

The assumed math is `20 × 2 + 100 × 1 = 140 credits`, within Free. Use Search for results first, then collect only approved pages. Keep citations, record publication dates separately from collection dates, and have an accountant verify tax claims. Firecrawl delivers source material; the firm writes the article and checks it. [Search](https://docs.firecrawl.dev/features/search), [prices](https://www.firecrawl.dev/pricing).

### Your own business listings across directories

For example, a cleaning company could inspect 30 directory pages about its own business four times a month. Permission basis: each directory permits this method, or supplies an approved API or export. Ownership of the business name does not confer ownership of the directory's website.

Collect public business name, address, hours, phone, and source link from ordinary HTML listings. Specialized data-provider products are excluded from this estimate. Skip reviewers' profiles and personal contact data. `30 × 4 × 1 = 120 credits`, within Free. Compare against the company's approved master record and send discrepancies to a person; do not let the collection job change listings. [Scrape](https://docs.firecrawl.dev/features/scrape), [prices](https://www.firecrawl.dev/pricing).

Run all five illustrations on one account and the total becomes `501 + 160 + 100 + 140 + 120 = 1,021 credits`. That exceeds Free by 21 credits and fits Hobby's 5,000 credits at $19 monthly. This is why each job needs a worksheet and a shared account budget. Switching 500 basic pages to JSON extraction instead would cost `500 × (1 + 4) = 2,500 credits`, before any other work.

## The permission checkpoint belongs before the API call

<figure><img src="/images/firecrawl-web-data-api-small-business-flow.svg" alt="Approved public page addresses pass through Firecrawl to text and source records for human review. Permission, access controls, and privacy or copyright questions determine whether collection proceeds or stops." loading="lazy" style="width:100%;height:auto"><figcaption>A small collection job starts with a permitted source and ends with a person checking what the data actually says.</figcaption></figure>

```text
Approved URLs → Firecrawl → Markdown + URL + time → Human review
                                                   ▼
                                           Assistant or report

Own site, written permission, or public use allowed by terms?
  No → Get written permission first
  Yes ▼
Login, paywall, CAPTCHA, IP block, or owner refusal?
  Yes → Stop and ask a lawyer
  No ▼
Personal data and copyright concerns resolved?
  No → Stop and review rights and privacy
  Yes → Go ahead with limits and logs
```

## Safe setup

The [README's Quick Start](https://github.com/firecrawl/firecrawl#quick-start) begins with an account and API key. Have a person create and approve the business account. The README also advertises a path for an agent to sign up its user; we recommend keeping signup, agreement acceptance, and payment decisions with the owner.

For a developer, an **SDK**, or software development kit, supplies convenient functions for making the API requests. These are the exact alternative commands in the [README's SDK sections](https://github.com/firecrawl/firecrawl#sdks), reproduced for an IT person to review. We did not execute them:

```bash
pip install firecrawl
```

```bash
npm install firecrawl
```

An **MCP server** is a program that hands an AI assistant a set of tools, like a contractor bringing a van of equipment for one job. The [local MCP documentation](https://docs.firecrawl.dev/mcp-server/local) requires Node.js 22 or later and npm/npx for local setup. Node.js runs JavaScript outside a web page; npm manages packages and npx launches them. The main README does not give minimum versions for every language SDK.

This is the [README's MCP configuration](https://github.com/firecrawl/firecrawl#mcp), with its placeholder key. Have IT put the real credential into a private configuration or secret manager, never into a public repository, browser page, screenshot, or shared kit:

```json
{
  "mcpServers": {
    "firecrawl-mcp": {
      "command": "npx",
      "args": ["-y", "firecrawl-mcp"],
      "env": {
        "FIRECRAWL_API_KEY": "fc-YOUR_API_KEY"
      }
    }
  }
}
```

This vendor example can download and launch a package when used. IT should inspect and pin an approved version for a business deployment. It is not a guarantee that every exposed tool is read only. Keep interaction actions and open ended Agent tasks outside the initial pilot. The README lists n8n and Zapier integrations; those workflow services have their own accounts, data routes, and costs.

Give IT the approved URL list, purpose, retention period, spend ceiling, and a named reviewer. Require a small explicit Crawl `limit`, `allowExternalLinks: false`, `allowSubdomains: false`, and `ignoreRobotsTxt: false`. The [Crawl reference](https://docs.firecrawl.dev/api-reference/endpoint/crawl-post) documents those controls and a `delay` between scrapes. Agree the pace with the owner; a maximum of one page a second is our conservative starting proposal, not a legal safe harbor.

The [Scrape reference](https://docs.firecrawl.dev/api-reference/endpoint/scrape) documents an automatic move to enhanced access when ordinary collection fails. Ask the provider how to prevent escalation on your approved workflow. Do not assume an assistant instruction will override server behavior. If you cannot enforce the owner's restrictions, use their approved feed or another collection method. Never retry through a different identity or network to defeat a denial.

Hosted requests and collected content pass through Firecrawl's servers. Its [privacy policy](https://www.firecrawl.dev/privacy-policy) says its servers are in the United States, lists client-collected information among the data it handles, and describes caching and indexing. It says personally identifying information is retained until written deletion is requested, subject to the policy. Do not promise customers that a basic plan discards their data immediately. The [Scrape reference](https://docs.firecrawl.dev/api-reference/endpoint/scrape) provides `storeInCache: false`, but that switch is not a complete retention agreement. Enterprise's advertised zero-data-retention arrangement requires its own terms. Your AI assistant and workflow provider may receive another copy.

## Self hosting costs and the license boundary

The [self-host guide](https://docs.firecrawl.dev/contributing/self-host) describes a Docker Compose evaluation setup. Docker packages services with their dependencies; Compose starts the related services together. The guide warns that this is not a finished production deployment: authentication, durable storage, encrypted connections, recovery, and operational support need work. Cloud features are not all available locally, and Agent can require Firecrawl Cloud or external services.

An existing server avoids a new rental bill only if it has spare capacity. Electricity, backups, updates, and the person handling failures still cost something. As a comparison, [Amazon Lightsail's Linux/Unix public IPv4 table](https://aws.amazon.com/lightsail/pricing/) listed a 4 GB memory, 2 virtual processor, 80 GB disk server at $24 per month when read at 21:37 UTC. That is a rental example, not a tested Firecrawl size or a capacity guarantee. Budget `server + backups + optional external services + IT hours × your actual hourly cost`; the guide does not establish a universal minimum or a predictable labor bill.

The repository's [root license](https://github.com/firecrawl/firecrawl/blob/main/LICENSE) is AGPL-3.0, covering the core unless a component says otherwise. We verified separate MIT licenses for the [Python SDK](https://github.com/firecrawl/firecrawl/blob/main/apps/python-sdk/LICENSE), [JavaScript SDK](https://github.com/firecrawl/firecrawl/blob/main/apps/js-sdk/LICENSE), [Elixir SDK](https://github.com/firecrawl/firecrawl/blob/main/apps/elixir-sdk/LICENSE), and [Ruby SDK](https://github.com/firecrawl/firecrawl/blob/main/apps/ruby-sdk/LICENSE). MIT permits commercial reuse subject to preserving its notices. The README also mentions some MIT UI components; do not apply that exception to unspecified core files.

Under [GNU AGPL section 13](https://www.gnu.org/licenses/agpl-3.0.en.html#section13), if you modify the covered program and let users interact with that version over a network, you must prominently offer those users its corresponding source at no charge. Building covered code into a product can also trigger source and distribution obligations; a separate MIT client calling the hosted API does not by itself turn your whole business application into AGPL code. Have a license reviewer examine the actual integration.

## The Computer Fraud and Abuse Act and other legal risks

This section is informational, not legal advice. Most ordinary small business uses can be lawful when confined to systems you own, public pages used within their terms, or work covered by written permission, with privacy and copyright addressed. Ask a lawyer licensed in your state about disputed access or commercial reuse before building a dependency on it.

The federal **Computer Fraud and Abuse Act**, or CFAA, is [18 U.S.C. § 1030](https://www.law.cornell.edu/uscode/text/18/1030). Section (a)(2) uses the phrase "without authorization or exceeds authorized access" when addressing intentional access and obtaining information; (a)(2)(C) concerns a protected computer. The definition in (e)(2)(B) reaches ordinary systems involved in interstate or foreign commerce or communication. Collecting a public page is not automatically a federal computer crime.

In **Van Buren v. United States**, 593 U.S. 374 (2021), the Supreme Court limited the exceeds-authorized-access theory: misusing information you were entitled to obtain is different from entering an area you were not entitled to access. Its "gates-up-or-down inquiry" appears on [opinion page 13](https://www.supremecourt.gov/opinions/20pdf/19-783_k53l.pdf). Footnote 8 leaves open whether relevant limits must be technical or can also arise from contracts or policies. Neither every terms violation nor every successful request settles the CFAA question.

In **hiQ Labs v. LinkedIn**, 31 F.4th 1180 (9th Cir. 2022), the Ninth Circuit found serious questions about applying the CFAA's without-authorization provision to generally public pages. This was a preliminary injunction appeal, not a nationwide Supreme Court license to collect anything. See [opinion pages 39 to 40](https://cdn.ca9.uscourts.gov/datastore/opinions/2022/04/18/17-16783.pdf). Circuit precedent does not bind every federal court in the country.

The later history matters. The district court's [November 4, 2022 order, pages 14 and 18](https://storage.courtlistener.com/recap/gov.uscourts.cand.312704/gov.uscourts.cand.312704.404.0.pdf), found User Agreement breaches and granted partial summary judgment concerning contractors' false identities, while leaving disputed defenses to the scraping claim unresolved. [Proskauer's December 8 report](https://newmedialaw.proskauer.com/2022/12/08/hiq-and-linkedin-reach-proposed-settlement-in-landmark-scraping-case/) says the court then entered the parties' consent judgment and permanent injunction, including $500,000 against hiQ. Those settlement stipulations were not new precedential findings granting or prohibiting all public-page collection.

In **Facebook v. Power Ventures**, 844 F.3d 1058 (9th Cir. 2016), continued access after written revocation and attempts to get around IP barriers supported CFAA liability. Users' permission did not override Facebook's revocation. [Opinion pages 19 to 20](https://cdn.ca9.uscourts.gov/datastore/opinions/2016/12/09/13-17102.pdf) explain the distinction; footnote 5 says an IP block alone would not suffice without more. Our operational rule is deliberately simple: a login, paywall, CAPTCHA, bot check, IP block, or refusal letter is the owner's no. Stop and resolve it. That rule is not a claim that every challenge conclusively proves a crime.

Damage and civil claims also matter. Section 1030(a)(5) addresses damaging systems, and (e)(8) includes impairment of availability. Excessive collection can hurt a small website. Section (g) permits qualifying civil actions; the $5,000 aggregate loss factor in (c)(4)(A)(i)(I) is one qualifying route, not a universal amount below which access is lawful. [Section (e)(11)](https://www.law.cornell.edu/uscode/text/18/1030) includes response and restoration costs.

### Contracts on both sides of the request

The target's terms can impose enforceable obligations independently of the CFAA. Firecrawl has its own agreement too. The [README disclaimer](https://github.com/firecrawl/firecrawl#license) says: "It is the sole responsibility of end users to respect websites' policies when scraping." It separately says: "By default, Firecrawl respects robots.txt directives."

That default is a useful start. **Robots.txt** is a website's machine readable instruction file for crawlers. [RFC 9309, section 1](https://www.rfc-editor.org/rfc/rfc9309), explicitly says, "These rules are not a form of access authorization." Following it alone does not make collection, storage, or reuse lawful. Check the permitted purpose as well as the permitted page.

Firecrawl's [Terms of Service](https://www.firecrawl.dev/terms-of-service), revised October 7, 2026, prohibit in section 5.d.1: "Use the Services for any commercial purposes except as expressly authorized by Firecrawl". Section 5.d.4 prohibits use "in violation of any laws, regulations, or contractual provisions". Before business use, obtain an order form or written confirmation covering your commercial purpose; do not assume a pricing card resolves that wording. Section 7.g places compliance obligations and potentially substantial reimbursement obligations for third-party claims on the user. Section 5.d also restricts personal-data solicitation or disclosure and uses such as debt collection, certain background checks, and credit-reporting purposes.

We found the prohibited-use list in those terms, but did not independently verify a separately linked acceptable-use policy. The robots statement quoted above is from the README, not a robots clause we found in the service agreement. Ask for all applicable policies before acceptance.

### Privacy, copyright, and state laws still apply

Pages can contain personal information. The [California attorney general's CCPA overview](https://oag.ca.gov/privacy/ccpa) describes rights and duties for covered businesses, with exclusions for defined publicly available information. It does not make every small business or every public business fact subject to the same duties. Have counsel check coverage, purpose, disclosures, retention, deletion, and your vendors; collect only the fields you need.

Converting copyrighted text or photographs into another format does not create reuse rights. [17 U.S.C. § 106](https://www.law.cornell.edu/uscode/text/17/106) identifies exclusive rights, while [§ 107](https://www.law.cornell.edu/uscode/text/17/107) sets out the fact-specific fair-use factors. A citation alone is not permission to republish a whole page. The AGPL obligations for covered software, discussed above, are separate from rights in collected content.

States have their own computer crime laws. [California Penal Code § 502(c)(2)](https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?sectionNum=502.&lawCode=PEN) addresses knowingly accessing and taking, copying, or using computer data without permission. [Texas Penal Code § 33.02(a)](https://statutes.capitol.texas.gov/?artSec=33.02&chapter=PE.33&code=PE&tab=2) addresses knowing access without the owner's effective consent. Other countries differ. Before starting, put the target URLs, allowed fields, collection method, pace, reuse purpose, retention period, signer, and stop conditions in writing.

## Take the starter kit to your IT person

- [Firecrawl web data starter kit (.md)](/downloads/firecrawl-web-data-starter-kit.md): the setup example, a credit worksheet, target permission record, request email, small business use cards, and a self-hosting license checklist.

For someone turning this into a paid research or support service, [*The $20 Dollar Agency*](/books/) fits the practical question: define the useful work a client buys, including who checks the result and who carries the operating responsibility.

## Related reading

- [The GitHub tools guide](/blog/github-top-repos-2026-small-business/)
- [Rea and reverse engineering risks](/blog/rea-reverse-engineering-agent-small-business-legal-risks/)
- [MoneyPrinterTurbo and small business video](/blog/moneyprinterturbo-ai-short-videos-small-business/)
- [Cloudflare's security audit skill](/blog/cloudflare-security-audit-skill-small-business/)
- [Scrapling and permission for web data](/blog/scrapling-web-data-small-business-legal-risks/)
- [MCP servers for small business](/blog/mcp-servers-for-small-business/)

## Fact-check notes and sources

All times below are UTC on November 3, 2026. These are documentation and selected-source readings, not a test of Firecrawl's results or claimed coverage. For the California statute and settlement report, we reviewed excerpts supplied from an ordinary browser at 21:39.

- [Live README](https://github.com/firecrawl/firecrawl#readme), 21:34: features, SDK commands, MCP example, account signup, agent onboarding, integrations, licensing statement, and responsibility disclaimer. [Selected API schema](https://github.com/firecrawl/firecrawl/blob/c0669540825571ef680c18f3e112af0bd6c4fb3c/apps/api/src/controllers/v2/types.ts#L1388), 21:37: defaults cross-check. This was a limited source read.
- [Pricing](https://www.firecrawl.dev/pricing), 21:34 and browser 21:36; [billing](https://docs.firecrawl.dev/billing), 21:35: plan prices, credits, concurrency, add-ons, rollover, and overage controls. Pricing and documentation disagree on some labels; those uncertainties are identified above. All workload quantities and arithmetic are illustrations.
- [Scrape](https://docs.firecrawl.dev/features/scrape), [Crawl](https://docs.firecrawl.dev/features/crawl), [Map](https://docs.firecrawl.dev/features/map), [Search](https://docs.firecrawl.dev/features/search), [Agent](https://docs.firecrawl.dev/features/agent), and [change tracking](https://docs.firecrawl.dev/features/change-tracking), 21:35: endpoint behavior and unit costs. [Monitor](https://docs.firecrawl.dev/features/monitoring), 21:35: scheduled checks and judging charges. [Interact](https://docs.firecrawl.dev/features/interact), 21:43: code and prompt billing, minimum session charge, initial scrape charged separately.
- [Crawl reference](https://docs.firecrawl.dev/api-reference/endpoint/crawl-post), [Scrape reference](https://docs.firecrawl.dev/api-reference/endpoint/scrape), and [local MCP](https://docs.firecrawl.dev/mcp-server/local), 21:35: scope settings, routing behavior, caching, and runtime requirements. [Self hosting](https://docs.firecrawl.dev/contributing/self-host), 21:34 to 21:35: evaluation limitations and operational work.
- [Terms](https://www.firecrawl.dev/terms-of-service) and [privacy policy](https://www.firecrawl.dev/privacy-policy), 21:35: commercial-use wording, prohibited uses, user obligations, data location, and retention. No separate acceptable-use document was verified.
- [Root license](https://github.com/firecrawl/firecrawl/blob/main/LICENSE), 21:34; [Python](https://github.com/firecrawl/firecrawl/blob/main/apps/python-sdk/LICENSE), [JavaScript](https://github.com/firecrawl/firecrawl/blob/main/apps/js-sdk/LICENSE), [Elixir](https://github.com/firecrawl/firecrawl/blob/main/apps/elixir-sdk/LICENSE), and [Ruby](https://github.com/firecrawl/firecrawl/blob/main/apps/ruby-sdk/LICENSE) licenses, 21:35: AGPL core and verified MIT exceptions. [GNU AGPL](https://www.gnu.org/licenses/agpl-3.0.en.html#section13), 21:35: section 13 and distribution obligations. [Lightsail](https://aws.amazon.com/lightsail/pricing/), 21:37: server rental comparison only.
- [CFAA](https://www.law.cornell.edu/uscode/text/18/1030), [Van Buren](https://www.supremecourt.gov/opinions/20pdf/19-783_k53l.pdf), and [hiQ appeal](https://cdn.ca9.uscourts.gov/datastore/opinions/2022/04/18/17-16783.pdf), 21:34: statutory scope and qualified holdings. [Power Ventures](https://cdn.ca9.uscourts.gov/datastore/opinions/2016/12/09/13-17102.pdf) and [hiQ contract order](https://storage.courtlistener.com/recap/gov.uscourts.cand.312704/gov.uscourts.cand.312704.404.0.pdf), 21:34: revocation and later contract history. [Settlement report](https://newmedialaw.proskauer.com/2022/12/08/hiq-and-linkedin-reach-proposed-settlement-in-landmark-scraping-case/), supplied excerpt reviewed 21:39: consent judgment; not an independent reading of the consent PDF.
- [RFC 9309](https://www.rfc-editor.org/rfc/rfc9309) and [CCPA overview](https://oag.ca.gov/privacy/ccpa), 21:34; [copyright § 106](https://www.law.cornell.edu/uscode/text/17/106), 21:34, and [§ 107](https://www.law.cornell.edu/uscode/text/17/107), 21:35; [Texas statute](https://statutes.capitol.texas.gov/?artSec=33.02&chapter=PE.33&code=PE&tab=2), browser 21:37; [California statute](https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?sectionNum=502.&lawCode=PEN), supplied excerpt reviewed 21:39: protocol, privacy, copyright, and state-law cautions.

*This post is informational, not legal advice. Mentions of third parties are nominative fair use. No affiliation or endorsement is implied.*


---

Canonical HTML: https://jwatte.com/blog/firecrawl-web-data-api-small-business/
RSS: https://jwatte.com/feed.xml
JSON Feed: https://jwatte.com/feed.json
Hero image: https://jwatte.com/images/firecrawl-web-data-api-small-business.webp