Desktop AI agent for Excel that edits the workbooks you already have — and gives them back intact. Tauri + Rust, surgical .xlsx patching, bring your own Claude/ChatGPT.
See the code
Let AI edit the Excel files you already have: faster, for fewer tokens, and you get them back intact.
GridPath is an engine for editing real .xlsx files. Give it to your coding
agent and the same edit takes a fraction of the time and tokens, because the
agent calls spreadsheet tools instead of writing and debugging scripts:
Claude Code (Sonnet 5.5) on a real 1.4 MB, 55,000-formula analyst model, same prompt with and without GridPath:
| Task | With GridPath | Without |
|---|---|---|
| Insert a row (median of 5 runs) | 17 s · 1.0k output tokens · $0.35 | 52 s · 6.1k · $0.86 |
| Change 12 inputs and report the recalculated result (1 run) | 29 s · 2.2k output tokens · $0.49 | 161 s · 17.1k · $1.62 |
Both lanes got every edit right. The difference is time, tokens, and seeing the diff before it's saved. Full runs, including the ones we lose, are in the evals repo.
It comes as two front doors on the same engine:
MCP server — npx gridpath mcp | Desktop app | |
|---|---|---|
| The agent | Yours: Claude Code, Cursor, Codex, any MCP host | GridPath's own, in the app |
| You review | In a local browser tab the agent links you to | In the app's diff panel |
| Install | One line, Node only | Download for Mac / Windows |
| Where | packages/gridpath · gridpath.dev/mcp | Releases · gridpath.dev |
Source-available under the FSL-1.1-Apache-2.0; each release becomes Apache 2.0 after two years.
claude mcp add gridpath -- npx -y gridpath mcp --review-required
Your agent edits the real .xlsx through GridPath's tools (describe, find,
read, write, run_script, structure ops). Every change returns a readback and a
review link: changed cells before/after, accept or reject per batch, save.
Nothing is written until you save; with --review-required only the review page
can save. No Python, no LibreOffice, no Excel needed. Details, host config
snippets and the tool list: packages/gridpath/README.md.
Open a .xlsx, type what you want, review the diff, accept it. Multiple
workbooks side by side, each with its own agent; other files attachable as
read-only references. Your file never leaves your machine, and it works with
the Claude or ChatGPT subscription you already pay for.
.xlsx stays on disk. Only your prompt and the cells the agent needs go to Claude or OpenAI.| GridPath | Excel Copilot | Claude for Excel | ChatGPT upload | CLI coding agent | |
|---|---|---|---|---|---|
Edits your existing .xlsx in place | ✅ | ✅ | ✅ | ❌ | ✅ |
| Workbook stays on your machine | ✅ | ❌ | ❌ | ❌ | ✅ |
| Preserves charts, pivots, plugin data | ✅ | ✅ | ✅ | ❌ | ❌ |
| Accept/reject diff before changes land | ✅ | partial | partial | ❌ | ❌ |
| Choose and switch the model | ✅ | ❌ | ❌ | ❌ | ✅ |
| Parallel sessions across workbooks | ✅ | ❌ | ❌ | ❌ | partial |
| Source available | ✅ | ❌ | ❌ | ❌ | ❌ |
| Included with a subscription you have | ✅ | ❌ | ✅ | ✅ | ✅ |
Best-effort comparison as of September 2026. The "preserves" and "diff" rows for coding agents are measured, not inferred — see the benchmark repo. On the August 2026 models a headless coding agent silently dropped parts of a workbook in 2 of 5 edit runs; on Sonnet 5.5 (September 2026) it preserved every part in all of our runs by choosing XML surgery on its own. GridPath preserves them by construction, not by the model's judgment on the day.
Under active development. Editing existing workbooks — the case above — is the strongest path and the one under continuous benchmark. Building large multi-sheet models from scratch works but is improving; the benchmark repo publishes those results too, including the ones we lose.
Source-available under the Functional Source License, Version 1.1, with Apache 2.0 Future License (FSL-1.1-Apache-2.0):
If you want to use GridPath in a way the FSL doesn't permit, get in touch.
Requires Node 20+, Rust stable via rustup, and your platform's Tauri prerequisites.
npm install
npm run tauri dev
npm run tauri build produces a local unsigned build. On first launch, connect Claude with OAuth or an API key, or use the ChatGPT sign-in flow. Credentials stay in the local SQLite database.
GridPath uses Tauri 2 with a Rust core, React and TypeScript, a vendored IronCalc fork (MIT/Apache-2.0), and SQLite. The surgical .xlsx patcher is in src-tauri/src/engine/workbook/xlsx_patch/, the agent loop in src-tauri/src/engine/spreadsheet_agent/, and the benchmark harness in eval/.
TypeScript
56.7%
Rust
28.0%
HTML
8.9%
JavaScript
2.1%
PLpgSQL
1.6%
CSS
1.3%
Desktop AI agent for Excel that edits the workbooks you already have — and gives them back intact. Tauri + Rust, surgical .xlsx patching, bring your own Claude/ChatGPT.
See the code
Let AI edit the Excel files you already have: faster, for fewer tokens, and you get them back intact.
GridPath is an engine for editing real .xlsx files. Give it to your coding
agent and the same edit takes a fraction of the time and tokens, because the
agent calls spreadsheet tools instead of writing and debugging scripts:
Claude Code (Sonnet 5.5) on a real 1.4 MB, 55,000-formula analyst model, same prompt with and without GridPath:
| Task | With GridPath | Without |
|---|---|---|
| Insert a row (median of 5 runs) | 17 s · 1.0k output tokens · $0.35 | 52 s · 6.1k · $0.86 |
| Change 12 inputs and report the recalculated result (1 run) | 29 s · 2.2k output tokens · $0.49 | 161 s · 17.1k · $1.62 |
Both lanes got every edit right. The difference is time, tokens, and seeing the diff before it's saved. Full runs, including the ones we lose, are in the evals repo.
It comes as two front doors on the same engine:
MCP server — npx gridpath mcp | Desktop app | |
|---|---|---|
| The agent | Yours: Claude Code, Cursor, Codex, any MCP host | GridPath's own, in the app |
| You review | In a local browser tab the agent links you to | In the app's diff panel |
| Install | One line, Node only | Download for Mac / Windows |
| Where | packages/gridpath · gridpath.dev/mcp | Releases · gridpath.dev |
Source-available under the FSL-1.1-Apache-2.0; each release becomes Apache 2.0 after two years.
claude mcp add gridpath -- npx -y gridpath mcp --review-required
Your agent edits the real .xlsx through GridPath's tools (describe, find,
read, write, run_script, structure ops). Every change returns a readback and a
review link: changed cells before/after, accept or reject per batch, save.
Nothing is written until you save; with --review-required only the review page
can save. No Python, no LibreOffice, no Excel needed. Details, host config
snippets and the tool list: packages/gridpath/README.md.
Open a .xlsx, type what you want, review the diff, accept it. Multiple
workbooks side by side, each with its own agent; other files attachable as
read-only references. Your file never leaves your machine, and it works with
the Claude or ChatGPT subscription you already pay for.
.xlsx stays on disk. Only your prompt and the cells the agent needs go to Claude or OpenAI.| GridPath | Excel Copilot | Claude for Excel | ChatGPT upload | CLI coding agent | |
|---|---|---|---|---|---|
Edits your existing .xlsx in place | ✅ | ✅ | ✅ | ❌ | ✅ |
| Workbook stays on your machine | ✅ | ❌ | ❌ | ❌ | ✅ |
| Preserves charts, pivots, plugin data | ✅ | ✅ | ✅ | ❌ | ❌ |
| Accept/reject diff before changes land | ✅ | partial | partial | ❌ | ❌ |
| Choose and switch the model | ✅ | ❌ | ❌ | ❌ | ✅ |
| Parallel sessions across workbooks | ✅ | ❌ | ❌ | ❌ | partial |
| Source available | ✅ | ❌ | ❌ | ❌ | ❌ |
| Included with a subscription you have | ✅ | ❌ | ✅ | ✅ | ✅ |
Best-effort comparison as of September 2026. The "preserves" and "diff" rows for coding agents are measured, not inferred — see the benchmark repo. On the August 2026 models a headless coding agent silently dropped parts of a workbook in 2 of 5 edit runs; on Sonnet 5.5 (September 2026) it preserved every part in all of our runs by choosing XML surgery on its own. GridPath preserves them by construction, not by the model's judgment on the day.
Under active development. Editing existing workbooks — the case above — is the strongest path and the one under continuous benchmark. Building large multi-sheet models from scratch works but is improving; the benchmark repo publishes those results too, including the ones we lose.
Source-available under the Functional Source License, Version 1.1, with Apache 2.0 Future License (FSL-1.1-Apache-2.0):
If you want to use GridPath in a way the FSL doesn't permit, get in touch.
Requires Node 20+, Rust stable via rustup, and your platform's Tauri prerequisites.
npm install
npm run tauri dev
npm run tauri build produces a local unsigned build. On first launch, connect Claude with OAuth or an API key, or use the ChatGPT sign-in flow. Credentials stay in the local SQLite database.
GridPath uses Tauri 2 with a Rust core, React and TypeScript, a vendored IronCalc fork (MIT/Apache-2.0), and SQLite. The surgical .xlsx patcher is in src-tauri/src/engine/workbook/xlsx_patch/, the agent loop in src-tauri/src/engine/spreadsheet_agent/, and the benchmark harness in eval/.
TypeScript
56.7%
Rust
28.0%
HTML
8.9%
JavaScript
2.1%
PLpgSQL
1.6%
CSS
1.3%