tidy-byte-tools
Provides 15 local MCP tools for repairing, converting, extracting, and checking text/files.
Repair broken or incomplete JSON from text or local files, including JSON Lines.
Turn key-value text, CSV/TSV, or Markdown tables into typed JSON from text or files.
Encode/decode Base64 for text or local files, with optional data URIs and URL-safe output.
Hash text or files with SHA-256, SHA-1, MD5, and SHA-512.
Split Markdown into RAG-ready chunks, list headings, or get one specific section.
Save Markdown chunks to a JSONL file.
Convert local PDF or HTML to Markdown, optionally writing the output to a file.
Extract PDF tables as CSV text or saved CSV files.
Check up to 200 web links for status codes, redirects, and errors.
Provides tools to convert PDFs or HTML files into Markdown, outline and retrieve Markdown sections, and split Markdown into chunks for RAG or search indexes.
Tidy Byte Tools
One install for our free local tools: fix broken JSON, pull tables out of a PDF, turn documents into Markdown, and more. It also starts one MCP server with all of those tools, so an AI assistant can use them on your computer. No API key.
This does not replace the individual lite repos. It is a doorway that installs them together.
Try it
This command is not valid JSON (the keys have no quotes, and there is a trailing comma):
printf '{name: "Ada", city: "Austin",}' | tidy-byte-tools json-repair -q-q hides a one-line tip about the paid cloud version. Real output from that command (0.1.0, 17 Nov 2026):
{
"name": "Ada",
"city": "Austin"
}
Related MCP server: ellmos-filecommander-mcp
Install
Needs Python 3.10 or newer. Install it from PyPI:
pip install tidy-byte-toolsOr use pipx, which keeps it in its own space:
pipx install tidy-byte-toolsThat also installs the eight lite tools. All of them come from PyPI. (hash-lite's PyPI name is tidy-hash-lite.)
Use it
printf '{name: "Ada", city: "Austin",}' | tidy-byte-tools json-repair
tidy-byte-tools pdf-table-pull report.pdf
tidy-byte-tools doc2md page.html -o page.md
tidy-byte-tools --helptbt is a short name for the same command. Installing the package also gives you the original commands (json-repair-lite, doc2md-lite, and so on).
From Python, import a lite package directly:
from json_repair_lite import repair_one
repair_one("{name: 'Ada',}")["value"] # {'name': 'Ada'}Use it from Claude Desktop, Cursor, or VS Code
The MCP server talks over stdio. After you install the package, add one of these configs. If you would rather not install it first, uvx can run it straight from PyPI.
Claude Desktop (claude_desktop_config.json) and Cursor (.cursor/mcp.json):
{
"mcpServers": {
"tidy-byte-tools": {
"command": "uvx",
"args": [
"--from",
"tidy-byte-tools",
"tidy-byte-tools-mcp"
]
}
}
}If the package is already installed, you can use "command": "tidy-byte-tools-mcp" and leave out args.
VS Code (.vscode/mcp.json):
{
"servers": {
"tidy-byte-tools": {
"command": "uvx",
"args": [
"--from",
"tidy-byte-tools",
"tidy-byte-tools-mcp"
]
}
}
}The server exposes 15 tools. Base64 tools are named base64_convert_text and base64_convert_file so they are not confused with the document converter. Most tools only read. save_chunks_jsonl, convert_to_markdown (when you set an output path), and extract_pdf_tables can write the files you ask for. check_links requests the URLs you give it.
You can also start the server yourself with tidy-byte-tools mcp.
The tools in this install
Command | What it does | Lite repo |
| Repair broken or incomplete JSON, including output from language models | |
| Turn notes, CSV/TSV, or Markdown tables into typed JSON | |
| Encode or decode Base64 for text and local files | |
| Hash text or files with SHA-256, SHA-1, MD5, and SHA-512 | |
| Split Markdown into chunks by heading, for RAG | |
| Turn a PDF or HTML file into clean Markdown | |
| Pull tables out of a PDF into CSV | |
| Check web links for status codes and redirects |
Paid cloud tools and the API
If you want the same kind of work in the cloud, pay per use:
Apify store: https://apify.com/tidybytetools
The site is https://tidybytetools.github.io.
Limits
These tools run on your computer. They are not a hosted service.
link-checkneeds a network. The other tools work offline.The PDF tools read text PDFs. A scanned page with no text layer will not give useful tables or Markdown.
This wrapper does not change how each lite tool works. See that tool's README for its flags and limits.
In the Official MCP Registry as
io.github.tidybytetools/tidy-byte-tools. You can also install the latest code from GitHub:pip install git+https://github.com/tidybytetools/tidy-byte-tools.
License
MIT. See LICENSE. Changes and ideas are welcome — see CONTRIBUTING.md.
Available Tools
15 toolsbase64_convert_fileEncode or decode a local file as Base64ARead-onlyIdempotent
Read a local file and encode its bytes to Base64, or decode a Base64 text file.
Use this when the data is in a file on this computer, so you don't have to paste it. The file is only read, never changed. Files larger than 5 MB are rejected; use the Apify version for bigger jobs. For decode, the file must be UTF-8 text containing Base64 (or a data URI), not a binary file.
path: full path to a file, up to 5 MB.
Returns the same result object as convert_text. If the decoded bytes are not UTF-8 text, "isText" is false and "text" is null.
Inputs: path (required), action, data_uri, url_safe.
base64_convert_textEncode or decode Base64 textARead-onlyIdempotent
Encode pasted text to Base64, or decode Base64 (or a data URI) back to text.
Use this when you have text or a Base64 string in the conversation. It is plain code, not an AI model, so the same input always gives the same output and nothing is made up. Do not use it for a file on disk (use convert_file for files), and do not use auto on short or unusual strings if you already know the direction — pass encode or decode instead.
Returns a result object. Encoded results include "action", "base64", "bytesIn", and "dataUri" plus "mime" when data_uri is true. Decoded results include "action", "bytesOut", "isText", and "text" (UTF-8 text, or null if the bytes are not text).
Inputs: text (required), action, data_uri, url_safe.
check_linksCheck linksARead-onlyIdempotent
Check web links and follow their redirects.
Use this when you have a list of URLs and need to know which ones work, which are broken (404, 500, timeouts, bad hosts) and where redirects end up, for example to clean up docs, a README, a sitemap list or bookmarks. It does not crawl a site or read page content; pass the exact links you want checked. Links that are not http or https come back with ok false and an error.
urls: up to 200 http(s) links. timeout: seconds to wait for each link.
Returns one result per link: {url, status, ok, final_url, redirects, error}. status is the HTTP code of the last hop (null if it could not connect). ok is true for 200-399. redirects lists each hop as "CODE url".
Inputs: urls (required), timeout. Returns: result
chunk_markdownSplit Markdown into chunksARead-onlyIdempotent
Split a Markdown file or Markdown text into chunks by heading, for search, embeddings or RAG.
Use this when you need a long Markdown document (docs, notes, a README) broken into pieces that each know where they came from. A chunk never mixes two sections. Each chunk has heading_path (like ["Guide", "Install"]), anchor (the GitHub-style #link), start_line and end_line (1-based, in the file), part and parts (when a big section had to be split), chars and text. Code blocks and tables are kept whole when they fit; when they don't, each piece gets the fence or table header again. Plain code, not AI: same input, same chunks.
Do not use it for PDF or Word files (convert them to Markdown first), to just see the headings (use outline_markdown), or to read one section (use get_section). Sizes are characters, not tokens.
Returns {"source", "total", "start", "chunks": [...], "next_start"}. next_start is null when there are no more chunks.
Inputs: path, text, limit, start, +3 more.
convert_to_markdownConvert PDF or HTML to MarkdownADestructiveIdempotent
Turn a local text PDF or saved HTML file into clean Markdown.
Use this when you want to read, quote, summarize or index a document as text, for example before RAG chunking or pasting into a prompt. Tables are kept as Markdown tables. Do not use it for scanned PDFs or photos (there is no OCR), for live web URLs (save the page as .html first), or when you only need the tables as CSV data (use a PDF table extractor instead).
path: full path to a .pdf or .html file on this computer. tables: PDF tables. "lines" (default) finds tables with border lines, "text" also tries tables lined up by spaces, "off" gives plain text. output_path: optional. If set, the Markdown is saved there and that path is returned. An existing file at that path is replaced. If empty, nothing is written and the Markdown text is returned.
Tables come back as Markdown tables. Scanned (image-only) PDF pages have no text and are skipped.
Inputs: path (required), tables, output_path. Returns: result
extract_pdf_tablesExtract PDF tables to CSVADestructiveIdempotent
Pull the tables out of a local text PDF.
Use this when you need the rows and columns of a table from a PDF on this computer, for example to load them into a spreadsheet, pandas or a database. Do not use it for scanned PDFs or photos (there is no OCR), for PDFs on the web (download the file first), or when you want the whole document as text (use a PDF-to-Markdown or text tool instead).
path: full path to a PDF on this computer. pages: optional, like "1,3-5". Empty means all pages. strategy: "auto" (default, tries borders first, then spacing), "lines" (tables with ruled borders) or "text" (tables lined up by spacing). Try "text" if "auto" misses a table that has no border lines. output_dir: optional. If set, each table is saved there as a CSV and the file paths are returned. Files named pageN_tableM.csv that already exist there are replaced. If empty, nothing is written and the CSV text of each table is returned.
Scanned (image-only) pages have no text and are skipped. Returns [] if no tables.
Inputs: path (required), pages, strategy, output_dir. Returns: result
get_sectionGet one Markdown sectionARead-onlyIdempotent
Return the text of one section of a Markdown file or text, found by heading title or anchor.
Use this when you only need one part of a long document, like the "Install" section of a README, so you don't have to read the whole file. If two headings have the same title, the first one is returned and "matches" tells you how many there are; use the anchor (from outline_markdown) to pick another.
Do not use it to split the whole document (use chunk_markdown).
Returns {"title", "heading_path", "level", "anchor", "start_line", "end_line", "chars", "matches", "text"}. If nothing matches, the error lists the headings.
Inputs: path, text, anchor, heading, +1 more.
hash_fileHash a local fileARead-onlyIdempotent
Hash a local file's raw bytes with SHA-256, SHA-1, MD5, and optional SHA-512.
Use this when the data is in a file on this computer, so you don't have to paste it. The file is only read, never changed. Files larger than 5 MB are rejected; use the Apify version for bigger jobs. This tool does not look the hash up in a malware list or a password list.
path: full path to a file, up to 5 MB. algorithms: a list from sha256, sha1, md5, sha512. Default is sha256, sha1 and md5. MD5 and SHA-1 are not safe for new security work; prefer SHA-256 or SHA-512 when you can.
Returns {"ok", "bytes", "hashes", and a hex field for each algorithm, plus a Base64 field when include_base64 is true}.
Inputs: path (required), algorithms, include_base64.
hash_textHash a text stringARead-onlyIdempotent
Hash pasted text with SHA-256, SHA-1, MD5, and optional SHA-512.
Use this when you have text in the conversation and need its checksum. The text is hashed as UTF-8 bytes. It is plain code, not an AI model, so the same text always gives the same hashes. This tool does not look the hash up in a malware list or a password list. Do not use it for a file on disk (use hash_file for files).
algorithms: a list from sha256, sha1, md5, sha512. Default is sha256, sha1 and md5. MD5 and SHA-1 are not safe for new security work; prefer SHA-256 or SHA-512 when you can.
Returns {"ok", "bytes", "encoding": "utf-8", "hashes", and a hex field for each algorithm, plus a Base64 field when include_base64 is true}.
Inputs: text (required), algorithms, include_base64.
outline_markdownList Markdown headingsARead-onlyIdempotent
List every heading in a Markdown file or text, with its level, path, anchor, line number and size.
Use this when you want a quick map of a long document before you read or chunk it, or to find the exact heading or anchor to pass to get_section. Headings inside code blocks and YAML front matter are ignored. Both "# Title" and underlined (=== / ---) headings are found.
Do not use it to get the text itself (use get_section or chunk_markdown).
Returns {"source", "count", "headings": [{"level", "title", "path", "anchor", "line", "end_line", "chars"}]}. chars is the size of that heading's own text, not counting the sections under it.
Inputs: path, text.
repair_json_fileRepair JSON in a local fileARead-onlyIdempotent
Read a local JSON file and repair it if it is broken or incomplete.
Use this when the JSON is in a file on this computer, so you don't have to paste it. The file is only read, never changed. Files larger than 5 MB are rejected; use the Apify version for bigger jobs. Do not use it for files that are not JSON text.
path: full path to a UTF-8 text file, up to 5 MB. json_lines: if true, each non-empty line is a separate document. If false (default), the whole text is tried as one JSON value first.
Returns the same object as repair_json_text.
Inputs: path (required), json_lines.
repair_json_textRepair broken JSON textARead-onlyIdempotent
Repair broken or incomplete JSON, including output from language models.
Use this when you have JSON in the conversation that may be missing quotes, have trailing commas, or be cut off. It is plain code, not an AI model: it can close missing brackets and fix common syntax issues, but it cannot invent the data you meant to write. Already-valid JSON is returned as-is, with wasValid true and repaired false. Do not use it for a file on disk (use repair_json_file for files).
json_lines: if true, each non-empty line is a separate document. If false (default), the whole text is tried as one JSON value first. If that cannot be repaired and there are several lines, each line is then tried on its own.
For one document, returns {"wasValid", "repaired", "value", "text"}. For several documents, returns {"count", "failed", "documents": [...]}.
Inputs: text (required), json_lines.
save_chunks_jsonlSave chunks to a JSONL fileADestructiveIdempotent
Split a Markdown file into chunks and save them as a JSONL file (one chunk per line).
Use this when the chunks are going into a search index, vector database or another program, and you want a file instead of a long chat reply. The chunks are the same as chunk_markdown gives. It writes only the one .jsonl file and replaces it if it is already there. It never changes the Markdown file.
Do not use it to look at chunks in the chat (use chunk_markdown).
Returns {"output_path", "chunks", "bytes", "biggest_chunk_chars"}.
Inputs: path (required), max_chars, output_path, prefix_path, +1 more.
shape_file_to_jsonTurn a text file into JSONARead-onlyIdempotent
Read a local text file (key: value notes, .csv, .tsv or a Markdown table) and return typed JSON.
Use this when the data is in a file on this computer, so you don't have to paste it. The file is only read, never changed. Do not use it for Excel (.xlsx), PDF or other binary files, for files that are already JSON, or for app logs.
path: full path to a UTF-8 text file, up to 5 MB. format: "auto" (default) picks for you. Or force "kv" (key: value or key=value lines, a blank line starts a new record), "csv" (comma, semicolon, tab or | separated, first row is the header) or "markdown" (a | pipe | table, first row is the header).
Returns {"format": , "data": }, the same as shape_text_to_json.
Inputs: path (required), format.
shape_text_to_jsonTurn text into JSONARead-onlyIdempotent
Turn pasted text (key: value notes, CSV/TSV or a Markdown table) into typed JSON.
Use this when you have plain-text records or a table in the conversation and need clean JSON you can trust, without guessing values yourself. It is plain code, not an AI model, so the same text always gives the same JSON and nothing is made up. Do not use it when the text is already JSON, is nested data, is app logs, or is a file on disk (use shape_file_to_json for files).
format: "auto" (default) picks for you. Or force "kv" (key: value or key=value lines, a blank line starts a new record), "csv" (comma, semicolon, tab or | separated, first row is the header) or "markdown" (a | pipe | table, first row is the header).
Returns {"format": , "data": }. data is a list of records for CSV and tables. For key: value text it is one object, or a list when blank lines split it into several records. Keys become lower_snake_case. Numbers become numbers, yes/no and true/false become booleans, empty or "null" becomes null.
Inputs: text (required), format.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 15 tools
Every tool targets a distinct input kind and outcome, and the descriptions explicitly cross-reference siblings with 'do not use X, use Y' guidance (text vs file variants, outline vs get_section vs chunk_markdown, convert_to_markdown vs extract_pdf_tables). The text/file pairs are the closest overlaps, but they are cleanly separated by where the data lives.
Most names follow a predictable verb_noun or resource_action pattern (repair_json_text, shape_file_to_json, hash_file, chunk_markdown, save_chunks_jsonl, check_links). The base64_convert_text / base64_convert_file pair uses a prefix-first ordering that breaks the pattern slightly, but the set remains highly readable.
Fifteen tools sits at the top of the well-scoped range and each one covers a distinct operation within a coherent local-file/plain-code utility domain. No tool looks redundant or filler.
The surface covers nearly every family symmetrically (text and file variants for repair, shaping, base64, hashing; full Markdown navigate/chunk/export lifecycle; PDF conversion and table extraction; link checking). Minor gaps exist, such as writing shaped or repaired JSON back to a file, or comparing/verifying hashes, but these are workable around.
Maintenance
Related MCP Connectors
MCP server for developer documentation, generated by doc2mcp.
MCP registry & directory: search, find & install 31k+ MCP servers & tools. Catalog and marketplace.
One MCP server for all your MCP servers: code mode, per-tool permissions, approvals and audit.
Related MCP Servers
- MIT
- AlicenseBqualityBmaintenanceComprehensive MCP server for filesystem operations, process management, interactive sessions, and async file search. Includes utilities for JSON repair, encoding fixes, duplicate detection, OCR, ZIP archives, and Markdown export.50417 npm6MIT
- AlicenseNot gradedqualityDmaintenanceProduction-grade MCP server for universal reverse engineering automation.85GPL 3.0
- FlicenseNot gradedqualityDmaintenanceA production-ready MCP server providing file, system, math, and text utilities through a simple CLI client.-