Firecrawl MCP Server

firecrawl_parse

Read-only

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. Authenticated final responses can include a data.metadata.scrapeId for optional parse feedback.

Input Schema

TableJSON Schema

Name	Required	Description
`proxy`	No
`maxAge`	No	Ignored: parse never reuses or stores indexed content.
`formats`	No
`parsers`	No
`filePath`	No	Phase 1 only: path to the local file on the caller/harness machine. Hosted MCP will not read or stat this path; it is used only to produce upload instructions.
`redactPII`	No
`uploadRef`	No	Phase 2 only: short-lived upload reference returned by phase 1 after the local PUT upload completes.
`pdfOptions`	No
`contentType`	No	Phase 1 MIME type override. If omitted, the server infers it from the file extension without reading the file.
`excludeTags`	No
`includeTags`	No
`jsonOptions`	No
`queryOptions`	No
`storeInCache`	No
`onlyMainContent`	No
`declaredSizeBytes`	No	Optional phase 1 size declaration. Hosted MCP does not stat the file; provide this only if the caller already knows it.
`zeroDataRetention`	No
`removeBase64Images`	No
`skipTlsVerification`	No

Output Schema

TableJSON Schema

Name	Required	Description
`raw`	No	Response body when it was not JSON.
`data`	No	Parsed document content; can include `data.metadata.scrapeId` for parse feedback.
`mode`	No	Which phase of the hosted flow produced this response.
`error`	No	Error message or error object when the call did not succeed.
`notes`	No	Hosted phase one: constraints on completing the upload flow.
`upload`	No	Hosted phase one: how to upload the local file.
`message`	No	Guidance for the next call.
`success`	No	Whether the API call succeeded.
`warning`	No	Non-fatal warning about the result.
`agent_hints`	No	Optional response guidance from the Firecrawl API.
`nextToolCall`	No	Hosted phase one: the second `firecrawl_parse` call to make once the upload succeeds, as `{name, arguments}`.

Schema Changelog

Changes observed during successful MCP inspections.

First observedSep 29, 2026

TDQS

A4.5/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already establish a safe read-only, non-destructive, closed-world operation, and the description adds substantial context on top: the two-phase upload behavior, what each call returns, PII redaction semantics, the zeroDataRetention account requirement, and the optional scrapeId feedback hook. None of this is derivable from 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.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with purpose and formats, then mechanics, then caveats, with no filler sentences. The one return-value sentence is arguably redundant given the output schema, but it is short and useful for disambiguating the two-phase flow.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 19-parameter tool with nested objects, the description nails the invocation flow and a few key flags but omits guidance on the options that govern the requested output modes (json/query schemas, tag filtering, formats/parsers). The presence of an output schema excuses return-value detail, but the parameter gaps keep this at minimum-viable.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

With only 26% schema description coverage across 19 parameters, the description must compensate more than it does. It meaningfully explains redactPII, zeroDataRetention, pdfOptions.maxPages, filePath and uploadRef, but leaves format-driving options (jsonOptions, queryOptions, includeTags/excludeTags, formats, parsers) undocumentable in either place.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Parse one supported document') and enumerates the output formats and the supported input file types. It also distinguishes itself from the sibling firecrawl_scrape by explicitly routing remote web URLs there.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives explicit operating conditions: local MCP reads filePath from the server filesystem, hosted MCP requires a two-call upload handshake with the warning not to send both fields together, and remote URLs belong in firecrawl_scrape. It also states the auth precondition for zeroDataRetention and tells anonymous users to omit it.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Firecrawl MCP Server tools Inspect

Try in Browser

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

Glama MCP Gateway

Add one secure layer between your agents and this server.