criptogus/HermesOffice

The world's first full-featured open-source AI Office suite.

Download · Privacy · Demo

HermesOffice is a free, open-source alternative to Microsoft Office for macOS, Windows, and Linux, built around AI editing as a first-class workflow rather than a bolted-on chat box. It opens and saves the real Microsoft Office formats — Word (.docx), Excel (.xlsx), PowerPoint (.pptx) — and edits PDF and Markdown too: a word processor, spreadsheet, presentation editor, PDF editor, and Markdown editor as six Electron apps sharing one engine layer.

Watch the demo video on YouTube

• Real PDF editing — retype text and edit images in the page itself, original fonts preserved.
• Local PDF → Word / PowerPoint / Excel conversion — turn a PDF into an editable .docx, .pptx, or .xlsx entirely on your machine: no cloud, no upload.
• Scanned PDFs too — on macOS and Windows scanned pages are read with the system OCR, so they convert to editable text.
• Microsoft Word–compatible, byte-preserving .docx editing — only what you touched changes; Word never notices.
• Word-faithful pagination — page breaks land where Word puts them.
• Excel-compatible spreadsheets — in-house engine with a Rust .xlsx sidecar, own charts, pivot tables, slicers.
• PowerPoint-compatible presentations — in-house .pptx engine with masters, layouts, smart guides, non-destructive crop.
• Markdown to Word, fully local — the same OOXML engine, no Pandoc, no cloud.
• AI that edits documents — block-level edits with snapshots and diffs, document-aware agents.
• Bring your own key (BYOK) — run the AI on your own API key: Claude, OpenAI, Gemini, DeepSeek, Kimi, GLM, Qwen, Doubao, MiniMax, Grok, Mistral, OpenRouter, or any OpenAI-compatible endpoint — or sign in with Genspark and skip keys entirely.
• Agent tools built in — web/image search, image generation, media analysis.
• Light / dark / system themes.
• macOS, Windows, Linux.
• Free & open-source (Apache-2.0).

macOS · Apple Silicon — download the latest .dmg One click, always the newest build. That link never changes — it is also the one to share or put on a page.

Builds come from main. The macOS bundle is ad-hoc signed, not notarized (there is no Apple Developer ID behind it yet), so the first launch needs one confirmation: right-click the app → Open, or allow it under System Settings → Privacy & Security.

Older versions are on the Releases page.

The deb installs with apt — it pulls in the dependencies and adds HermesOffice to the applications menu:

sudo apt install ./hermesoffice_<version>_amd64.deb

On Fedora / RHEL-family / openSUSE, install the rpm instead:

sudo dnf install ./hermesoffice-<version>.x86_64.rpm     # Fedora / RHEL family
sudo zypper install ./hermesoffice-<version>.x86_64.rpm  # openSUSE

The AppImage instead runs in place: install the FUSE 2 runtime (sudo apt install libfuse2; on Ubuntu 24.04 the package is libfuse2t64), make the file executable, then run it:

chmod +x HermesOffice-<version>.AppImage
./HermesOffice-<version>.AppImage

Every app embeds the same AI panel: block-granular AI editing with version snapshots and diffs in docs, a tool-calling agent over workbook/slide/PDF state in the others.

The whole suite ships light / dark / system UI themes built on shared design tokens (packages/ui), with a CI guard that keeps chrome colors on the token system. Document surfaces stay light in dark mode — Word-style dark chrome around white paper — so files render and export identically in both themes.

AI backends — Genspark sign-in or bring your own key. By default the apps sign in to a Genspark account through a device-code flow — no model API key to enter — and model calls route through the Genspark proxy (Claude, GPT, and Gemini families). Or bring your own key (BYOK) in the AI settings: Claude, OpenAI, Gemini, DeepSeek, Kimi, GLM, Qwen, Doubao, MiniMax, Grok, Mistral, and OpenRouter are built in, plus a custom provider slot for any OpenAI-compatible endpoint (base URL + key), local servers included. A Genspark account also unlocks the Genspark ("gsk") tool endpoints the agents build on — web and image search, image generation and editing, image/audio/video analysis, and audio transcription — all reachable through packages/ai-search for anyone extending the agent layer.

All pure TypeScript, no Electron dependency, unit-tested (except the UI kit):

• packages/docx-engine — docx parsing → block tree (with docxIndex anchors and passthrough), OOXML fragment generation, byte-level paragraph patching.
• packages/pptx-engine / packages/pptx-render — pptx model and rendering.
• packages/pdf2docx — local PDF → DOCX conversion: PDFium character-level extraction, pure-geometry layout analysis, rebuild through docx-engine; the same analysis drives the PDF app's PowerPoint and Excel exports.
• packages/file-parse — text extraction for AI attachments (office formats, text formats).
• packages/agent-core — the AI agent loop and skill composition shared by every app.
• packages/ai-provider — provider abstraction and streaming for the model backends.
• packages/ai-search — Genspark auth + web/image search tools.
• packages/i18n, packages/ui, packages/project-store, packages/electron-utils — shared i18n core, React UI kit, recent-files store, and Electron main-process helpers.

npm install
npm run fixtures     # generate test .docx fixtures
npm test             # engine + app unit tests (docs/sheets/slides need no display)
npm run typecheck    # tsc --noEmit across every workspace
npm run dev          # all five editors + shell against Vite dev servers
npm run dev:docs     # a single app (same pattern works per workspace)
npm run dist:mac     # package macOS dmg (regenerates third-party notices)
npm run dist:win     # package Windows nsis installer
npm run dist:linux   # package Linux AppImage + deb + rpm

The sheets app additionally needs a Rust toolchain for its xlsx sidecar (cargo on PATH); npm run build -w @hermesoffice/sheets compiles it automatically.

Local UI/e2e driver scripts (Playwright + Electron, for local acceptance, not committed by default) live in scripts/drivers/.

open docx ─► archive original by hash (never touched)
          ─► docx-engine parses word/document.xml top-level elements (w:p / w:tbl / …)
          ─► Block tree, each block anchored by docxIndex + original XML slice
          ─► Tiptap streaming editor (manual + AI editing, dirty tracking)
save      ─► dirty blocks → OOXML fragments (referencing existing styles only)
          ─► splice into original document.xml (untouched blocks keep original bytes)
          ─► repack zip; all other entries copied byte-for-byte

The same philosophy holds in sheets and slides: the original file is the source of truth, edits are applied as narrow patches, and everything the editor didn't touch survives the round trip untouched.

Is HermesOffice free? Yes. HermesOffice is free and open-source under the Apache-2.0 license — no trial, no paid tier for the apps themselves.

Can HermesOffice open Microsoft Word, Excel, and PowerPoint files? Yes. HermesOffice opens and saves native .docx, .xlsx, and .pptx files. Saving is byte-preserving: parts of the file you didn't touch are written back byte-for-byte, so documents keep working in Microsoft Office.

Does HermesOffice work offline? Document editing is fully local — files never leave your machine to be opened, edited, or saved. The AI features (agents, search, image tools) need a network connection, with either a Genspark sign-in or your own model API key (BYOK).

Can HermesOffice edit PDF files? Yes — real PDF text and image editing that rewrites the page content stream with the original fonts preserved, not cover-up annotations.

Can HermesOffice convert PDF to Word, Excel, or PowerPoint? Yes — HermesOffice converts PDFs into editable .docx, .xlsx, and .pptx files entirely on-device: PDFium character-level extraction plus geometry-based layout analysis, no cloud service, no upload. Scanned pages are covered too — on macOS and Windows the system OCR reads them, so they convert to editable text rather than a page image.

Can I use my own AI model or API key? Yes. Besides the keyless Genspark sign-in, HermesOffice supports bring your own key (BYOK) for Claude, OpenAI, Gemini, DeepSeek, Kimi, GLM, Qwen, Doubao, MiniMax, Grok, Mistral, and OpenRouter, plus any OpenAI-compatible endpoint — including local model servers.

Does HermesOffice collect any data? Official packaged builds send limited usage analytics by default, and you can disable reporting at any time under Settings → General. Analytics never sends document content, file names, file paths, account identity, or email addresses. See HermesOffice Privacy for the complete event and data disclosures.

See SECURITY.md for the process security posture (renderer sandboxing, IPC validation, external-link gating) and the threat models for AI-generated content.

HermesOffice would not be possible without these open-source projects:

• Electron — the desktop runtime for every app.
• Univer (Apache-2.0) — the spreadsheet UI core that Sheets extends.
• PDFium (BSD-3-Clause, bundled via @embedpdf/pdfium) — the content-stream engine behind true PDF text and image editing.
• pdf.js (Apache-2.0) and pdf-lib (MIT) — PDF rendering and document assembly.
• Tiptap / ProseMirror — the block editors in Docs and Markdown.
• Konva — canvas rendering for Slides and Sheets charts.
• HarfBuzz (wasm) — text-shaping metrics for complex scripts.
• calamine and IronCalc — the read and calc layers of the Rust xlsx sidecar.
• Liberation, Carlito, Caladea, and Noto CJK fonts (OFL/Apache-2.0) — bundled document fonts.

npm run notices regenerates the bundled third-party license summary (tools/gen-third-party-notices.mjs); all runtime dependencies are MIT/Apache-2.0/BSD-3-Clause/OFL, and the bundled fonts (Liberation, Carlito, Caladea, Noto CJK subsets) are OFL/Apache.

HermesOffice is licensed under the Apache License 2.0, with one exception: the ee/ directory is reserved for future enterprise modules and is covered by the HermesOffice Enterprise License.

The HermesOffice and Genspark names and logos are trademarks of Mainfunc, Inc. The Apache-2.0 license does not grant permission to use them (see section 6); forks should use their own branding.