Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Conv2pdf: convert and edit PDFs from your agent

    1

    Teaches your agent to use the Conv2pdf MCP server well: files and web pages to PDF, PDF to Word or images, merge, split, number, compress and protect, at the least cost in quota.

    Free

    1 useSecurity scanned
    Conv2pdf: convert and edit PDFs from your agent

    Conv2pdf: convert and edit PDFs from your agent

    Example session with this skill installed

    Merge the three invoices in ~/Desktop/tests/agensi-demo/invoices into one PDF named Q3-invoices.pdf, number the pages, and protect it with the password Q3-accounts-2026.

    • Read your context and instructions
    • Compiled the conv2pdf

    Done: the three invoices are in one PDF of 4 pages, numbered and protected.

    Final file: ~/Desktop/tests/agensi-demo/invoices/Q3-invoices.pdf (8,181 bytes). It asks for the password Q3-accounts-2026 to open.

    What I did

    Your key is on the Dev plan, which merges 2 files per call, so the job took 4 conversions instead of 3. I protected last: a tool that edits a protected PDF writes a result without its password.

    StepToolInputResult
    1merge_pdfsinvoice-2026-07.pdf + invoice-2026-08.pdfinvoice-2026-07-merged.pdf
    2merge_pdfsthe result of step 1 + invoice-2026-09.pdfQ3-invoices-merged.pdf
    3add_page_numbersQ3-invoices-merged.pdfQ3-invoices-numbered.pdf, pages marked "1 / 4" to "4 / 4"
    4protect_pdfQ3-invoices-numbered.pdfQ3-invoices.pdf

    The invoices come in the order of their names: July, August (2 pages), September.

    Quota

    4 conversions used. The key was at 130 of 300 before, it is at 134 now: 166 left.

    Next steps

    • Your three original invoices are untouched.
    • Three intermediate files are still in the folder: invoice-2026-07-merged.pdf, Q3-invoices-merged.pdf and Q3-invoices-numbered.pdf. None of them has a password. Shall I delete them?
    • unlock_pdf cannot find a forgotten password: keep Q3-accounts-2026 somewhere safe.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    This skill tells an AI agent how to use the tools of the official Conv2pdf MCP server on the files of your computer. The server gives the agent 15 tools. The skill gives it the judgement: which tool to call, in which order, and what each call costs.

    What the agent learns

    • Which tool to call. Word, Excel, PowerPoint, OpenDocument, Pages, RTF, TXT and CSV files, PNG, JPG, WebP, GIF, TIFF and HEIC images, EPUB books, PostScript and HTML files to PDF, and a public web page from its address. PDF to Word or to images. Merge, split into one file per page, extract pages, rotate, number, watermark, compress, protect and unlock.
    • The order of a job in several steps. Merge, then extract or rotate, then number or watermark, then compress, and protect last, so that the password is not lost on the way.
    • What a call costs. One unit of your quota per conversion. The agent knows which refusals are free and which are counted, counts the calls before a long job, and checks the quota before a whole folder.
    • What to say and what to ask. Where the result was written, which files are intermediate, when a compression gained little, and never to guess a password.

    What you need

    • The Conv2pdf MCP server: npx -y conv2pdf-mcp, with Node.js 22 or later. Installation.
    • A Conv2pdf API key. The Dev plan is free: 300 conversions, valid for 12 months. conv2pdf.com/en/api.

    Privacy

    A file given to a tool is uploaded to Conv2pdf, hosted in the EU (France), and deleted once the result is on your disk. The skill itself makes no network call and runs no code: it is a single SKILL.md.

    Source

    MIT licence. Source and issues: github.com/conv2pdf/mcp.

    No OCR: a scanned PDF does not become a Word document. Numbers and Keynote files are not converted. Needs the Conv2pdf MCP server (Node.js 22 or later) and a Conv2pdf API key. Files are uploaded to Conv2pdf, hosted in the EU (France), and deleted once the result is downloaded. On the free Dev plan: 300 conversions in all, files of 10 MB at most, 2 files per merge call.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 21 days ago

    • Free with an Agensi account

    Requires

    Conv2pdf
    Node.js

    Permissions

    Read Files
    Write Files
    Network Access
    Environment Variables

    Allowed Hosts

    conv2pdf.com

    The skill is a single SKILL.md: it runs no code and makes no network call. The Conv2pdf MCP server it relies on reads the files you name, uploads them to Conv2pdf, hosted in the EU (France), where they are deleted once the result is downloaded, and writes the results next to them. Its API key is set in the configuration of the MCP server, never in the conversation.

    Listed1 month ago
    Updated15 days ago

    What's inside

    Frequently Asked Questions

    Converting Word, Excel, PowerPoint, OpenDocument, Pages, image, EPUB, PostScript and HTML files to PDF, as well as public web pages from their address, turning a PDF into a Word document or into images, and merging, splitting, extracting pages, rotating, numbering, watermarking, compressing, protecting and unlocking PDFs, for one file or for a whole folder.

    Any agent that loads SKILL.md skills and can run an MCP server: the tools come from the conv2pdf-mcp npm package, which needs Node.js 22 or later. In Claude Code, the Conv2pdf plugin installs the server and the skill together.

    Yes, a Conv2pdf API key: the conversions run on the Conv2pdf servers. The Dev plan is free and gives 300 conversions, valid for 12 months. You set the key in the configuration of the MCP server, not in the conversation.

    The agent gives a tool the path of a file. The file is uploaded to Conv2pdf, hosted in the EU (France), converted, and the result is written next to the original or where you ask. Nothing is overwritten, and Conv2pdf deletes its copy once the result is downloaded.

    No. The Conv2pdf MCP server has no OCR tool: a scanned PDF is refused by the PDF to Word conversion, and the skill tells the agent to say so instead of trying again.

    New in Productivity