Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.
Scored across 3 tools
Each tool targets a clearly distinct operation: parse handles local document files, scrape handles a single known URL, and search handles query-based discovery. The descriptions explicitly reinforce these boundaries, such as noting that remote web URLs belong in firecrawl_scrape rather than firecrawl_parse.
All three tools follow the same firecrawl_ prefix followed by a single action verb: parse, scrape, search. This creates a predictable and consistent naming convention with no mixed styles or vague verbs.
Three tools is within a reasonable range and each covers a distinct core capability for the server's parsing and extraction purpose. The count feels slightly lean for a Firecrawl integration, but it is not bloated or artificially padded.
The set covers document parsing, single-URL scraping, and web search, which handles many common workflows. However, the descriptions reference firecrawl_crawl, firecrawl_map, and firecrawl_research_* tools that are not available here, leaving notable multi-page and research-oriented gaps that agents must work around.
Parse one supported document into markdown, HTML, links, summary, targeted answers, or JSON matching a schema. Supported inputs include common HTML, PDF, Word, RTF, OpenDocument, and spreadsheet files; PDF parsing can be bounded with pdfOptions.maxPages.
Local MCP reads filePath from the server filesystem. Hosted MCP uses two calls: first provide filePath to receive upload instructions, upload locally, then call again with the returned uploadRef; do not send both fields together. Remote web URLs belong in firecrawl_scrape.
Set redactPII to request redaction of personally identifiable information in the returned content. zeroDataRetention requires an eligible authenticated account; omit it for anonymous keyless use. Returns upload instructions for hosted phase one or parsed document content for the final call.
| Name | Required | Description | Default |
|---|---|---|---|
| proxy | No | ||
| maxAge | No | ||
| formats | No | ||
| parsers | No | ||
| filePath | No | ||
| redactPII | No | ||
| uploadRef | No | ||
| pdfOptions | No | ||
| contentType | No | ||
| excludeTags | No | ||
| includeTags | No | ||
| jsonOptions | No | ||
| queryOptions | No | ||
| storeInCache | No | ||
| onlyMainContent | No | ||
| declaredSizeBytes | No | ||
| zeroDataRetention | No | ||
| removeBase64Images | No | ||
| skipTlsVerification | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and destructiveHint=false; description builds on this by disclosing the local vs hosted two-phase upload flow, pdfOptions.maxPages bounding, redactPII behavior, and the zeroDataRetention account requirement. No contradictions.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with the main purpose, then compactly covers the most important usage variations and options in four sentences. No repetition of schema or annotation data.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a complex 19-parameter tool with no output schema, the description covers core behavior, return value, local/hosted workflows, and exclusions. It does not mention several secondary options (e.g., includeTags/excludeTags, onlyMainContent, proxy, maxAge), but those are not central to selecting the tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description compensates for 0% schema coverage by explaining the local/hosted filePath/uploadRef relationship, pdfOptions.maxPages, redactPII, zeroDataRetention eligibility, and the output types (markdown/html/links/summary/query/json). It leaves some self-explanatory options (proxy, maxAge, tags) unprompted, but the semantically opaque parameters are mostly clarified.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb-object statement: 'Parse one supported document into markdown, HTML, links, summary, targeted answers, or JSON matching a schema.' It also distinguishes itself from sibling tools by explicitly stating 'Remote web URLs belong in firecrawl_scrape.'
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It explicitly says when to use the tool (supported local documents) and what not to use it for ('Remote web URLs belong in firecrawl_scrape'). It further gives usage flow for local vs hosted MCP, including the two-call upload instruction and the rule 'do not send both fields together.'
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Retrieve and extract content from one supplied URL through Firecrawl. Use this when the request identifies a page and needs its content or defined fields. It can return markdown, HTML, links, screenshots, branding data, a targeted answer, or JSON matching a supplied schema; JSON is useful when the requested result has defined fields, while markdown preserves readable page content.
This tool operates on a known page. For a set of pages use firecrawl_crawl, and to discover page URLs use firecrawl_map or firecrawl_search. Options include JavaScript render delay, cache age, main-content filtering, PII redaction, and lockdown cache-only retrieval. Browser actions may change the live page when interactive actions are enabled.
Firecrawl may reuse recently indexed content instead of refetching the page, and the reuse window varies by domain. Set maxAge: 0 to force a live fetch, or a smaller maxAge to bound how stale reused content may be. A successful response does not by itself confirm that the state it describes is still current.
Returns the selected content formats and page metadata.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | ||
| proxy | No | ||
| maxAge | No | ||
| mobile | No | ||
| formats | No | ||
| parsers | No | ||
| profile | No | ||
| waitFor | No | ||
| location | No | ||
| lockdown | No | ||
| redactPII | No | ||
| pdfOptions | No | ||
| excludeTags | No | ||
| includeTags | No | ||
| jsonOptions | No | ||
| queryOptions | No | ||
| storeInCache | No | ||
| onlyMainContent | No | ||
| screenshotOptions | No | ||
| zeroDataRetention | No | ||
| removeBase64Images | No | ||
| skipTlsVerification | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations declare readOnlyHint=true, but the description states 'Browser actions may change the live page when interactive actions are enabled,' implying a write/modification capability. This directly contradicts the read-only annotation. Additionally, the description does not reconcile this contradiction, leaving the agent uncertain about side effects. Per the rubric, a score of 1 is required due to the annotation contradiction.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is moderately long but well-structured into four logical paragraphs: purpose, alternatives, options, and caching caveats. It front-loads the primary purpose in the first sentence. Some redundancy exists (e.g., repeated mention of 'known page'), but for a tool with 22 parameters, the length is justified and every paragraph serves a distinct informational goal.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity (22 params, no output schema, sparse annotations), the description covers a lot: usage context, format selection guidance, alternative tools, caching behavior with maxAge, and a caveat about staleness. It also states what is returned ('selected content formats and page metadata'). It is not perfect—return value details are minimal and the readOnly contradiction undermines completeness—but overall it is rich and situationally useful.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0%, so the description must compensate for the 22 undocumented parameters. It does add meaningful semantics for key params like maxAge ('Set maxAge: 0 to force a live fetch'), formats (markdown vs JSON), and option categories like main-content filtering and PII redaction. However, many parameters (proxy, location, mobile, profile, screenshotOptions, etc.) remain unexplained, and the generic 'Options include' phrase does not fully cover the breadth. This is adequate but leaves clear gaps.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's function: 'Retrieve and extract content from one supplied URL through Firecrawl.' It names a specific verb-resource pair, differentiates from sibling tools by noting this operates on a known page while firecrawl_crawl handles sets of pages and firecrawl_map/firecrawl_search discover URLs, and enumerates concrete output formats (markdown, HTML, screenshots, JSON). This is specific and unambiguous.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly states when to use the tool: 'Use this when the request identifies a page and needs its content or defined fields.' It also provides clear alternatives: for a set of pages use firecrawl_crawl, and for discovery use firecrawl_map or firecrawl_search. This gives the agent explicit decision criteria relative to sibling tools.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search web, news, or image sources and return ranked results. Operators include quoted phrases, -term, site:host, inurl:term, intitle:term, and related:host; the set is non-exhaustive. includeDomains and excludeDomains are mutually exclusive hostname filters; categories limit results to GitHub, research, PDF, or developer sources.
For a programming question, add categories: ["developer"]. It searches an index of repositories, GitHub issues, merged pull requests, repository READMEs, and curated documentation sites, and returns the hits in data.web with category: "developer".
categories: ["research"] restricts these web results to research-affiliated websites and returns page snippets. The firecrawl_research_* tools are a separate surface that searches paper abstracts and full text across biomedical (PubMed, bioRxiv, medRxiv) and arXiv literature.
Each web result is a title, URL, and description, not the page. Add scrapeOptions to attach page content in the same call; those fetches ignore maxAge, so use firecrawl_scrape when you need a live fetch. Returns source-type result groups and usage metadata. Authenticated responses can include an id for optional search feedback.
| Name | Required | Description | Default |
|---|---|---|---|
| tbs | No | ||
| limit | No | ||
| query | Yes | ||
| filter | No | ||
| sources | No | ||
| location | No | ||
| categories | No | ||
| enterprise | No | ||
| highlights | No | ||
| scrapeOptions | No | ||
| excludeDomains | No | ||
| includeDomains | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already mark read-only and non-destructive, and the description adds meaningful behavior: result snippets are not page content, scrapeOptions fetches bypass maxAge, developer category results land in data.web with category: 'developer', and authenticated responses may include an id. None of this contradicts the annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is longer than minimal but front-loads the core purpose in the first sentence and organizes topics into purpose, category guidance, and result/scrape behavior. It is dense and does not contain filler, though it could be trimmed around peripheral details such as the optional authenticated id.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with 12 parameters and no output schema, the description covers the main call flow, return shape (ranked title/URL/description groups), category behavior, and scrape-vs-scrapeOptions tradeoff. It omits guidance on several less-common search parameters, but is complete enough for correct basic invocation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description gives real meaning for query operators, includeDomains/excludeDomains mutual exclusivity, sources, categories, and scrapeOptions, which is important because schema description coverage is 0%. However, several parameters remain unexplained (tbs, filter, location, enterprise, highlights, limit), so the description only partially compensates for the schema's lack of descriptions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Opens with a specific verb and resource: 'Search web, news, or image sources and return ranked results.' It further distinguishes search results from page content by stating each result is a title, URL, and description, and points to firecrawl_scrape for live fetches. This is enough to separate it from the scrape/parse siblings.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Gives explicit when-to-use guidance: add categories: ['developer'] for programming questions, and use categories: ['research'] selectively. It names firecrawl_research_* as a separate surface for biomedical/arXiv literature and explicitly says to use firecrawl_scrape when a live fetch is needed because scrapeOptions fetches ignore maxAge.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
firecrawl_search3 fields changedInput schema / properties / limit / maximumAdded value: +100Input schema / properties / limit / minimumAdded value: +1Input schema / properties / limit / typePrevious value: -"number"New value: +"integer"firecrawl_search1 field changedInput schema / properties / categories / items / enumPrevious value: -[
- "github",
- "research",
- "pdf"
-]New value: +[
+ "github",
+ "research",
+ "pdf",
+ "developer"
+]firecrawl_agentfirecrawl_agent_statusfirecrawl_check_crawl_statusfirecrawl_crawlfirecrawl_extractfirecrawl_feedbackfirecrawl_interactfirecrawl_interact_stopfirecrawl_mapfirecrawl_monitor_checkfirecrawl_monitor_checksfirecrawl_monitor_createfirecrawl_monitor_deletefirecrawl_monitor_getfirecrawl_monitor_listfirecrawl_monitor_runfirecrawl_monitor_updatefirecrawl_research_inspect_paperfirecrawl_research_read_paperfirecrawl_research_related_papersfirecrawl_research_search_githubfirecrawl_research_search_papersfirecrawl_search_feedback
Add one secure layer between your agents and this server.
