BibTeX Reference Manager MCP
Offline, read-only MCP server for working with BibTeX, RIS, and CSL JSON reference lists without sending data anywhere.
Parse and normalize entries: field names, whitespace, DOIs, years, and authors.
Convert between BibTeX, RIS, CSL JSON, and plain-text citation styles (APA, MLA, IEEE, Chicago, Harvard).
Detect and remove duplicates by DOI, cite key, or similar titles; optionally show merged fields.
Lint entries for missing fields, bad page ranges, uncapitalized titles, invalid DOIs, odd years, repeated keys, and unescaped characters.
Apply safe fixes for page ranges, title capitalization, and bare DOIs.
Generate consistent cite keys like
smith2024deep, with letters for repeats.Return structured results and readable text while never modifying files or making network requests.
Provides automated tools to manage and fix academic bibliographies, specifically designed to lint, format, and maintain BibTeX files used in LaTeX documents.
Handles and cleans reference lists (BibTeX, RIS) exported from Mendeley, automating the process of linting, deduplication, and format conversion for use in academic writing.
Handles and cleans reference lists (BibTeX, RIS) exported from Zotero, automating the process of linting, deduplication, and format conversion for use in academic writing.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@BibTeX Reference Manager MCP Lint my references.bib and tell me what's broken in each entry."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
BibTeX Reference Manager MCP
BibTeX Reference Manager MCP is a free, open source MCP server, published on npm as bibtex-reference-manager-mcp. It cleans, converts, dedupes and formats BibTeX, RIS and CSL JSON reference lists, so you stop fixing broken .bib files by hand. Paste a list into your AI app and ask it to read the entries, convert them to another format, find the duplicates, flag the problems or give every entry a tidy cite key.
It is made with students in mind. Writing a thesis in LaTeX? Ask it to fix a broken .bib file, clean up a Zotero export or a Mendeley export, turn an EndNote RIS file into BibTeX, and tidy your thesis bibliography before you hand it in.
Use it with Claude Desktop, Claude Code, Cursor, VS Code Copilot and any other app that supports MCP. It runs on your computer, needs no internet and no login, and your references never leave your machine.
Free forever. No account and no subscription.
Install
Run it straight from npm with npx. You need Node.js 20 or newer, see Node version for the details. There is nothing else to install, no account, no API key and no config file for the server itself.
npx -y bibtex-reference-manager-mcpRelated MCP server: zotero-native-mcp
Try these prompts
Paste your reference list into the chat first, or attach the file, then ask:
"Here is my references.bib. Lint it and tell me what is broken, entry by entry."
"Find the duplicates in this list and give me back a clean BibTeX file without them."
"Convert these RIS entries from my library export to BibTeX, then rename the keys so they look like smith2024deep."
Add it to your AI app's MCP config so the app can start it for you:
All of them start the same server the same way. The only difference is where the config goes.
Claude Desktop
Add this to claude_desktop_config.json:
{
"mcpServers": {
"bibtex": {
"command": "npx",
"args": ["-y", "bibtex-reference-manager-mcp"]
}
}
}Restart Claude Desktop after you save the file.
Claude Code
Run this in a terminal:
claude mcp add bibtex -- npx -y bibtex-reference-manager-mcpCursor
Cursor reads .cursor/mcp.json from the project folder, or ~/.cursor/mcp.json in your home folder if you want the server in every project.
{
"mcpServers": {
"bibtex": {
"type": "stdio",
"command": "npx",
"args": ["-y", "bibtex-reference-manager-mcp"]
}
}
}VS Code Copilot
Add this to .vscode/mcp.json in your workspace:
{
"servers": {
"bibtex": {
"type": "stdio",
"command": "npx",
"args": ["-y", "bibtex-reference-manager-mcp"]
}
}
}Why this server
Some MCP servers connect to a reference manager account and read your library online. This one does not. It works on text you paste in, so it is a different tool for a different job.
It needs no account, no API key and no internet. Nothing is sent anywhere.
It does not connect to Zotero or Mendeley. You export a file and paste it in, so it works with any reference manager that can write BibTeX, RIS or CSL JSON.
The work is done by code, not by the AI guessing. Converting, deduping, linting and renaming keys give the same result every time.
It is read only. It never edits your library or your files. You get the result back and decide what to keep.
If you want an assistant to search or change a live online library, you need a server that connects to that service. This one is for cleaning up an exported list.
Competitor comparison
Other MCP servers work with reference libraries too. Most connect to an online account or a web API, so your data leaves your computer and you need an account or an API key. This server works on text you paste in, so it needs no account, no API key and no internet, and nothing is sent anywhere.
Server | Works offline | Needs account or key | Reads a live library | Read only | Formats |
BibTeX Reference Manager MCP | Yes | No | No | Yes | BibTeX, RIS, CSL JSON |
zotero-mcp | No | Yes | Yes | No | Zotero, BibTeX |
mendeley-mcp | No | Yes | Yes | No | Mendeley, BibTeX |
bibtex-mcp | No | Optional | No | Yes | BibTeX |
reference-mcp | No | No | No | Yes | BibTeX |
CrossRef-MCP | No | No | No | Yes | BibTeX |
mcp_doi2bib | No | No | No | Yes | BibTeX |
Local-Zotero-MCP | No | No | Yes | Yes | Zotero |
jabref-mcp | No | No | Yes | No | BibTeX |
zotero-mcp and mendeley-mcp connect to your online Zotero or Mendeley library and can change it. bibtex-mcp, reference-mcp, CrossRef-MCP and mcp_doi2bib fetch citations from web sources such as Semantic Scholar, OpenAlex, CiteAs, Google Scholar and CrossRef. Local-Zotero-MCP reads your local Zotero library through the Zotero desktop app. jabref-mcp reads a JabRef .bib file and can add entries through JabRef.
Pick this server when you want to clean, convert, dedupe, lint and rename an exported list without sending it anywhere. Pick one of the others when you want to search or change a live online library.
Export from Zotero or Mendeley
In Zotero, right click a collection (or select some items), choose the export option and pick BibTeX or RIS. In Mendeley, select your references, choose Export and pick BibTeX or RIS. Menu names change between versions, so look for Export.
Open the saved
.bibor.risfile in any text editor and copy everything, or attach the file in an app that accepts attachments.Paste it into your AI app and ask, for example: "Lint this list, remove the duplicates and give me back clean BibTeX with tidy cite keys."
Copy the result into your
.bibfile. To get it back into your reference manager, save it as a.bibor.risfile and import it.
The format is detected for you. If a file is not recognized, say which format it is.
Tools
Tool | What it does |
| Reads BibTeX, RIS or CSL JSON text and returns normalized entries: lower case field names, tidy spacing, bare DOIs, the year filled in from a date, and authors split into "Family, Given". |
| Converts between BibTeX, RIS, CSL JSON and plain text. The input format is detected unless you pass |
| Finds duplicates by DOI, by the same cite key and by similar titles (the years must not clash). Says which entry to keep and returns a BibTeX list without the extra copies. |
| Flags missing required fields, bad page ranges, uncapitalized titles, invalid DOIs, odd years, repeated keys and unescaped |
| Applies the safe fixes from |
| Makes keys from the first author, the year and the first useful title word, like |
Every tool takes the reference list as text, and every tool returns both a readable answer and structured fields. All of them are read only. They never change a file, because you paste the text in and get the result back.
Formats
BibTeX:
@stringmacros, quoted and braced values,#joins, and@commentblocks are handled.RIS: the usual tags such as
TY,AU,TI,T2,PY,SP,EP,DOandKW.CSL JSON: a list of items, one item, or an object with an
itemslist.
Common workflows
These show how the six tools fit together. All of them take the reference list as text and are read only.
Clean a Zotero or Mendeley export
Run
parse_bibliographyto normalize the entries and fill in missing years.Run
dedupe_entriesto remove duplicates by DOI, cite key or title.Run
lint_entriesto see missing fields, bad page ranges and odd years.Run
generate_cite_keysto give every entry a tidy key likesmith2024deep.Run
convert_bibliographyto get back a clean BibTeX file.
Turn an EndNote RIS export into BibTeX
Run
parse_bibliographyon the RIS text to check it parses.Run
convert_bibliographywithtoset to BibTeX.Run
generate_cite_keysif you want keys likesmith2024deep.
Check a finished .bib before you hand it in
Run
parse_bibliographyto normalize.Run
lint_entries, thenfix_entriesto apply the safe fixes it can make.Run
dedupe_entriesto make sure nothing is listed twice.
Privacy
Everything runs on your computer. The server reads only the text your AI app passes to it. It makes no network requests, does not look up DOIs online, needs no API key and keeps nothing between calls. DOI checks only look at the shape of the DOI, they do not ask a website whether it exists.
FAQ
Does it check that a DOI really exists? No. It runs offline, so it only checks that the DOI has the right shape, like 10.1000/xyz123.
How does it decide two titles are the same? It lowercases the titles, drops punctuation and accents, and compares pairs of letters. The default match is 0.9 out of 1. Pass title_threshold to loosen or tighten that. Two entries with different years are not matched by title.
Will it change my protected capitals like {DNA}? No. Braces are kept in BibTeX output and only removed when you convert to RIS, CSL JSON or plain text.
Node version
You need Node.js 20 or newer to run the server. To work on it, use Node.js 22.12 or newer.
Development
npm install
npm run build
npm testThe code is in src. src/bib.ts reads and writes the formats, src/analyze.ts finds duplicates and lint problems, and src/tools.ts defines the six tools.
More free student MCP servers
All free, open source and read only where they touch your accounts. Browse the full list at brightspacemcp.com/more-mcps.
Server | What it does |
D2L Brightspace for AI. Due dates with real submission status, grades, a target grade calculator and course files. | |
Canvas LMS for AI, read only. Due dates, missing work, grades and announcements. | |
Moodle for AI, read only. Due dates, overdue work, grades and announcements. | |
Google Classroom for AI, read only. Todos, due dates, missing work and grades. | |
GPA, grade conversion and the grade you need on the final, for AI. Works offline, no account. | |
APA, MLA and Chicago citations from a DOI, an ISBN or typed details. No account. | |
Exam study schedules, spaced repetition dates, pomodoro plans and assignment milestones. Works offline. | |
Search scholarly papers, find open access copies and related work through OpenAlex. No key needed. | |
Find free legal copies of textbooks on OpenStax, DOAB and Open Library, look up editions by ISBN and suggest cheaper older ones. | |
Chemistry and physics homework math: molar mass, balancing equations, grams to moles, unit conversion and significant figures. Works offline. | |
Offline writing metrics for essays. Word and page counts, readability scores, passive voice, repeated openers, filler words, long sentences and word limit checks, so students can check their work against assignment requirements. No account, no network. | |
Syllabus to calendar MCP server. Extract dates from syllabus text, build an ics calendar, expand recurring classes and merge calendars. Works offline, no account. | |
Job application tracker MCP server. Track internships, jobs, interviews and deadlines offline, see what is coming up, and export your list as CSV or Markdown. No account and no API keys. |
License
MIT, see LICENSE.
Available Tools
6 toolsconvert_bibliographyConvert a bibliographyARead-only
Convert a reference list between BibTeX, RIS, CSL JSON and plain text (a simple reference list style, APA by default, or MLA, IEEE, Chicago or Harvard when you pass style). The input format is detected unless you pass it. Page ranges, DOIs and names are cleaned on the way. Example: text is a BibTeX entry and to is "ris", and output is the same reference as RIS. Set from only to override detection.
Inputs: to (required), text (required), from, style. Returns: to, from, count, output
dedupe_entriesFind duplicate entriesARead-only
Find duplicate references by matching DOI, by the same cite key, and by similar titles (the years must not clash). Each group says which entry to keep, the one with the most fields, and which to remove. Also returns the list as BibTeX with the extra copies dropped. Example: two entries with the same DOI, one with 3 fields and one with 6, give one group that keeps the entry with 6 fields and lists the other under remove. Your files are not changed.
Inputs: text (required), format, title_threshold, merge_duplicates. Returns: total, groups, merged, unique, +1 more
fix_entriesFix entriesARead-only
Apply the safe fixes that lint_entries points to and return the cleaned list as BibTeX: page ranges with one hyphen get a double hyphen, titles that start with a lower case letter or are all lower case get capitals (words inside braces are left alone, and acronyms like DNA, NASA or iOS in those titles are wrapped as {DNA} so BibTeX keeps their case; they are listed in protected_titles), and DOIs written as a link or with a doi: prefix become bare DOIs. Missing fields, backwards page ranges and DOIs that are not valid are not guessed, so run lint_entries on the result. Example: pages "12-15" becomes "12--15". Your files are not changed.
Inputs: text (required), format. Returns: count, fixed, fixes, bibtex, +1 more
generate_cite_keysGenerate cite keysARead-only
Create consistent cite keys from the first author's family name, the year and the first meaningful word of the title, like smith2024deep. Repeated keys get a letter added. Returns the old and new key for each entry and the list as BibTeX with the new keys. Example: an entry with key old1, author "Smith, John", year 2024 and title "Deep Learning for Graphs" gives old_key old1 and new_key smith2024deep, or smith2024 when include_title_word is false.
Inputs: text (required), format, show_bibtex, include_title_word. Returns: keys, count, bibtex
lint_entriesLint entriesARead-only
Check a reference list for problems: missing required fields for the entry type, bad page ranges, titles that are not capitalized, invalid DOIs, odd years, repeated keys and unescaped &, %, _ or # in titles, journal names and other text fields (rule unescaped-character). Each issue names the entry key, the field and a plain message. Example: an @article with no journal field gives an issue with the rule missing-field and the field journal. Pass the whole list as text so repeated keys can be found.
Inputs: text (required), format. Returns: count, errors, issues, warnings
parse_bibliographyParse a bibliographyARead-only
Read BibTeX, RIS or CSL JSON text and return the entries in one normalized shape: lower case field names, tidy whitespace, bare DOIs, the year filled in from a date, and authors split into "Family, Given". Nothing is sent anywhere. Example: text "@article{k1, AUTHOR = {John Smith and Jane Doe}, title={Deep Learning for Graphs}, year={2024}}" returns one entry with authors ["Smith, John", "Doe, Jane"] and lower case field names. Leave format out unless auto picks the wrong one.
Inputs: text (required), format. Returns: count, format, entries
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
Each tool has a distinct role: parse normalizes, convert changes format, lint reports, fix applies safe repairs, dedupe groups duplicates, generate_cite_keys rewrites keys. The only mild overlap is parse_bibliography vs convert_bibliography, since both ingest and re-emit a list, but descriptions make the difference (normalized shape vs output format) reasonably clear.
Every tool follows a verb_object snake_case pattern (convert_bibliography, parse_bibliography, lint_entries, fix_entries, dedupe_entries, generate_cite_keys). No mixing of casing or style, and the object noun reflects the domain (entries vs bibliography).
Six tools is well-scoped for a reference-manager pipeline: ingest, validate, repair, deduplicate, rekey, and export. Each tool covers a distinct step with no redundant entries, and nothing feels padded.
The lifecycle of cleaning and converting a bibliography is covered (parse, lint, fix, dedupe, keys, convert) with a sensible text-only, no-file-write stance. Minor gaps exist: no merge/add-entry operation and no tool to look up or validate a DOI against a resolver, but agents can work around these with parse plus convert.
Maintenance
Related MCP Connectors
Turn DOIs, ISBNs, arXiv or PubMed ids into verified references and export them in any style.
Remote MCP for JSON repair, hashing, formatting, IDs, dates, units, and other small tools.
Remote MCP: JSON repair, JSON Schema validation, JSON to TypeScript, CSV to JSON and JSON to CSV.
Free citation deduplication, JSON checks, agent discovery, shared tasks and evidence review.
Related MCP Servers
- FlicenseNot gradedqualityCmaintenanceEnables academic integrity auditing, manuscript preparation, literature search, journal formatting, and submission packaging through local MCP tools.-
- AlicenseAqualityBmaintenanceMCP server for Zotero 10+. Read and write your library entirely offline through Zotero's native local API. No plugin, no API key, no cloud.28240 npm13MIT
- AlicenseNot gradedqualityCmaintenanceEnables MCP clients to audit and fix EPUB, Office, OpenDocument, SVG, and WACZ files entirely offline via the Lens-suite CLI tools, returning stable JSON results.MIT
- AlicenseBqualityCmaintenanceEnables locally-first literature research through an MCP host by collecting sources, parsing documents, retrieving evidence, validating citations, and exporting research records in multiple formats.38MIT