Agent Data Tools: Documents to text MCP server
io.github.Nutshell2000/agent-data-tools-documentsNutshell2000streamable-httpv1.331.0Scanned
5 tools: PDF, Word, Excel, PowerPoint and web pages to Markdown; invoice fields as JSON.
5 tools: 4 change things and 1 only read.
- Reads private data · 1
- Reads text others write · 5
- Can send data out · 5
- 1 finding
Tools listed without sign-in. Remote posture
With Anlyon
4 tools here change things. Govern them.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the Agent Data Tools: Documents to text MCP server can do
The Agent Data Tools: Documents to text MCP server is published by the GitHub account Nutshell2000. It has 5 tools. It can create or change data (4 tools: extract_rendered, pdf_text, document_text and document_extract), but cannot delete data. 1 tool only reads.
1 tool reads private data (extract); 5 tools bring in text other people can write (extract, extract_rendered, pdf_text and 2 more); 5 tools can send data out of your system (extract, extract_rendered, pdf_text and 2 more). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
The fixed checks matched 1 finding (annotation-conflict). Each one comes with the text it matched.
Its remote endpoint lists its tools without sign-in, and its tools changed 10 times since Anlyon Radar first saw them.
Last scanned Nov 21, 2026, version 1.331.0.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
5 tools
Changes things 4Claude Code: ask
document_extractExtract named fields from a document as JSON: give a PDF, Word, Excel or PowerPoint file, a scan or a photo (PNG, JPEG, WebP) by URL or in a POST body (up to 10 MB; images 3 MB) an…Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Extract named fields from a document as JSON: give a PDF, Word, Excel or PowerPoint file, a scan or a photo (PNG, JPEG, WebP) by URL or in a POST body (up to 10 MB; images 3 MB) and the field names you want, such as invoice_number, total, due_date. An AI model reads the file, scans by OCR, and returns one JSON object, with null where a value is not stated. Use to turn invoices, receipts, orders and contracts into rows. Price $0.03/call via x402 USDC or a card key (/credits); failed calls free. Reads or computes only and changes nothing; each call is charged, so repeats are not free.
Annotations the server declares
- title: Extract data from a document: invoice, receipt, contract fields as JSON
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
document_textConvert a document to text or Markdown: Word (.docx), Excel (.xlsx), PowerPoint (.pptx), PDF, plain UTF-8 text (.txt/.md/.csv/.json), and scans or images (PNG, JPEG, WebP) by OCR.Reads text others writeCan send data out1 finding
Why it is in this group
- The description's first verb is "convert": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Findings
- Warn
The server marks "document_text" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":true}; The description's first verb is "convert": a word for a change.
Description, as the server wrote it Convert a document to text or Markdown: Word (.docx), Excel (.xlsx), PowerPoint (.pptx), PDF, plain UTF-8 text (.txt/.md/.csv/.json), and scans or images (PNG, JPEG, WebP) by OCR. Give a public URL or POST the file (10 MB default; 10-20 MB URLs retry once at 20 MB; images 3 MB). Old formats and protected files fail uncharged. Use for attachments, scans and spreadsheets in an agent, RAG or workflow. Example: url=https://calibre-ebook.com/downloads/demos/demo.docx. Price $0.005/call via x402 USDC or a card key (/credits); failed calls free. Reads or computes only and changes nothing; each call is charged, so repeats are not free.
Annotations the server declares
- title: Document to text: Word, Excel, PowerPoint, PDF
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
extract_renderedSame output as /extract, but the page is first rendered in a real browser, so single-page apps and pages that build their content with JavaScript come back with their text instead …Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Same output as /extract, but the page is first rendered in a real browser, so single-page apps and pages that build their content with JavaScript come back with their text instead of an empty shell. Waits for the network to settle (up to 15 seconds). Public pages only; no login. Use when /extract returns EXTRACTION_EMPTY or too little text for a page you can read in a browser. Example: url=https://example.com/. Price $0.01/call via x402 USDC or a card key (/credits); failed calls free. Reads or computes only and changes nothing; each call is charged, so repeats are not free.
Annotations the server declares
- title: Web page to Markdown, rendered: a JavaScript-rendered page as text or Markdown
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
pdf_textExtract the text of a public PDF (up to 10 MB) from its URL, page by page, with title, author and page count.Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Extract the text of a public PDF (up to 10 MB) from its URL, page by page, with title, author and page count. Choose the first page, how many pages (up to 200) and maxChars. arXiv abstract links work. If a site blocks our fetch, the Internet Archive's dated copy is read and the answer says so. Scanned, encrypted and non-PDF files fail uncharged. Use to read papers, invoices, reports and manuals into an agent. Example: url=https://bitcoin.org/bitcoin.pdf. Price $0.005/call via x402 USDC or a card key (/credits), first 10 free; failed calls free. Reads or computes only and changes nothing; each call is charged, so repeats are not free.
Annotations the server declares
- title: PDF to text, PDF text extraction: extract the text of a PDF document from a URL, page by page
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
Only reads 1Show 1 read-only tool
extractFetch any public web page URL and get clean, LLM-ready text or Markdown: main content, code blocks, tables, links and headings, for RAG and agent context.Reads private dataReads text others writeCan send data outNo findings
Why it is in this group
- The description's first verb is "fetch": a word for reading.
What it can reach
- private-dataThe description's first sentence says "fetch" and "table": it reads databases.
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- untrusted-inputThe description's first sentence says "fetch" and "web": it reads web pages, which anyone can write.
- untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Claude Code: ask, because it is tagged outbound.
Description, as the server wrote it Fetch any public web page URL and get clean, LLM-ready text or Markdown: main content, code blocks, tables, links and headings, for RAG and agent context. Static HTML, Markdown or text; optional section ID. Default 20,000 output characters, max 100,000; input 2 MiB. No JavaScript rendering (see /extract/rendered), logins or paywalls. Empty pages, bad sections, browser challenges and errors are not charged. Use to read docs or articles. Example: url=https://example.com. Price $0.002/call via x402 USDC or a card key (/credits), first 10 free; failed calls free. Reads or computes only and changes nothing; each call is charged, so repeats are not free.
Annotations the server declares
- title: Extract web content, Markdown: any URL to clean text for LLMs
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
25 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 private-data tool breaks every one.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write5
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out5
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
{
"permissions": {
"deny": [
"mcp__agent-data-tools-documents__extract"
]
}
}Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 15, 2026, 13:23 UTC, unchanged since Nov 15, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- yes
- Tools that change things, listed without credentials
- 4
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- https://agent-data-tools.revmesh2074.workers.dev/.well-known/oauth-protected-resource
- Authorization servers
- https://agent-data-tools.revmesh2074.workers.dev
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- yes
- Client ID metadata documents
- no
- WarnThe whole server
The endpoint served tools/list without credentials, and the list has 4 tools that change things (4 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://agent-data-tools.revmesh2074.workers.dev/mcp/documents: tools/list returned 5 tools to an anonymous client, including extract_rendered (write), pdf_text (write), document_text (write), document_extract (write)
Findings
Matches of fixed checks, with the text they matched.
- Warndocument_text
The server marks "document_text" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":true}; The description's first verb is "convert": a word for a change.
History
Changes to its tool definitions and to its remote endpoint, newest first.
Tools4 changed.~4 changedView diff
- Changed:document_extractChanges thingsdescription
- Changed:document_textChanges thingsdescription
- Changed:extractOnly readsdescription
- Changed:pdf_textChanges thingsdescription
Tools1 changed.~1 changedView diff
- Changed:extractOnly readstitleannotations
Tools2 changed.~2 changedView diff
- Changed:document_textChanges thingsdescription
- Changed:pdf_textChanges thingsdescription
Tools1 changed.~1 changedView diff
- Changed:pdf_textChanges thingsinputSchema
Tools1 changed.~1 changedView diff
- Changed:document_textChanges thingsdescription
Tools1 changed.~1 changedView diff
- Changed:document_textChanges thingsdescriptioninputSchema
Tools1 changed.~1 changedView diff
- Changed:document_textChanges thingsdescriptioninputSchema
Tools1 changed.~1 changedView diff
- Changed:document_extractChanges thingsinputSchema
Tools1 changed.~1 changedView diff
- Changed:document_extractChanges thingsinputSchema
Tools2 changed.~2 changedView diff
- Changed:extractOnly readsinputSchema
- Changed:extract_renderedChanges thingsinputSchema
Install
Pinned to version 1.331.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http agent-data-tools-documents https://agent-data-tools.revmesh2074.workers.dev/mcp/documentsclaude mcp add --transport http agent-data-tools-documents https://agent-data-tools.revmesh2074.workers.dev/mcp/documentsRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [],
"ask": [
"mcp__agent-data-tools-documents__extract",
"mcp__agent-data-tools-documents__extract_rendered",
"mcp__agent-data-tools-documents__pdf_text",
"mcp__agent-data-tools-documents__document_text",
"mcp__agent-data-tools-documents__document_extract"
],
"deny": []
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as agent-data-tools-documents. Review it before you use it.
Capability tags behind 5 rules
- mcp__agent-data-tools-documents__extract private-data, untrusted-input, outbound
- mcp__agent-data-tools-documents__extract_rendered untrusted-input, outbound
- mcp__agent-data-tools-documents__pdf_text untrusted-input, outbound
- mcp__agent-data-tools-documents__document_text untrusted-input, outbound
- mcp__agent-data-tools-documents__document_extract untrusted-input, outbound
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/io.github.Nutshell2000/agent-data-tools-documents/verify?hash=dce44bdc51c2e8e1c3b8f8bccf2d46dbff2bebef75f87201f4dcf7e937b60296" | jq -e '.changed == false'Pinned hash dce44bdc51c2, from Nov 21, 2026, 19:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.Nutshell2000/agent-data-tools-documents/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.Nutshell2000/agent-data-tools-documents)Source: MCP Registry · repository. Synced Nov 21, 2026, 19:17 UTC.
Agent Data Tools: Documents to text MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Agent Data Tools: Documents to text MCP server safe?
- No scan can promise that. What its definitions show: 5 tools, 4 of which can change, send or delete something, and 1 finding from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the Agent Data Tools: Documents to text MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the Agent Data Tools: Documents to text MCP server do things that cannot be undone?
- No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.
What tools does the Agent Data Tools: Documents to text MCP server have?
- 5 tools: 4 write, 1 read. extract_rendered, pdf_text, document_text, document_extract and extract.
Does the Agent Data Tools: Documents to text MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the Agent Data Tools: Documents to text MCP server send data to outside services?
- 5 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): extract, extract_rendered, pdf_text, document_text and document_extract.
Has the Agent Data Tools: Documents to text MCP server changed its tools?
- Yes, 10 times since Anlyon Radar started tracking it, most recently on Nov 21, 2026. Every change is on this page and in its Atom feed.