OfficeCLI

OfficeCLI is an open-source, single-binary command-line tool built specifically for AI agents to create, read, modify, and automate Word (.docx), Excel (.xlsx), and PowerPoint (.pptx) files. It ships under the Apache 2.0 license and requires no Microsoft Office installation or external runtime dependencies. The project is maintained by iOfficeAI on GitHub and reached v1.0.129 as of July 2026.

What It Is

OfficeCLI is an AI-native Office automation CLI that replaces multi-library Python scripts with single commands. Where manipulating a PowerPoint slide previously required 50+ lines of python-pptx code, OfficeCLI reduces it to one command like officecli add deck.pptx / --type slide --prop title="Q4 Report". It operates on all three major Office formats with full read, modify, and create support, and exposes every operation as deterministic JSON output — making it straightforward for AI agents to parse results without regex scraping.

Three-Layer Architecture

OfficeCLI organizes its commands into three progressive layers so agents and developers can start simple and go deep only when needed:

• L1 (Read): Semantic views via view — modes include text, outline, annotated, stats, issues, html, svg, and screenshot
• L2 (DOM): Structured element operations — get, query, set, add, remove, move, swap — using stable path-based addressing like /slide[1]/shape[2]
• L3 (Raw XML): Direct XPath access via raw, raw-set, add-part, and validate — a universal fallback when L2 isn't sufficient

Built-in Engines

The binary ships several self-contained engines that require no external tools:

• Rendering engine: A from-scratch HTML renderer that converts .docx/.xlsx/.pptx to HTML or per-page PNG screenshots, enabling the render → look → fix loop in headless CI/Docker environments. The watch command starts a local HTTP server with auto-refresh on every edit.
• Formula & pivot engine: 350+ Excel functions auto-evaluated on write, including spilling dynamic arrays (FILTER, SORT, UNIQUE, LAMBDA), financial math (XIRR, PRICE, YIELD), and statistical functions. Native OOXML pivot tables can be created from a source range with a single command.
• Template merge: The merge command replaces {{key}} placeholders in any .docx/.xlsx/.pptx with JSON data across paragraphs, table cells, shapes, headers, footers, and chart titles.
• Round-trip dump: dump serializes any document or subtree into replayable batch JSON; batch replays it — enabling agents to learn from human-authored templates.

AI Agent Integration

OfficeCLI is designed to integrate directly with AI coding agents. Running officecli install auto-detects Claude Code, Cursor, Windsurf, GitHub Copilot, and similar tools and installs its skill file into each. A built-in MCP server can be registered with a single command (officecli mcp claude, officecli mcp cursor, etc.), exposing all document operations as JSON-RPC tools. The SKILL.md file at officecli.ai/SKILL.md teaches any agent how to install the binary and use all commands in one curl fetch. Structured error codes (not_found, invalid_value, unsupported_property) include suggestions and valid ranges so agents can self-correct without human intervention.

Installation and Platform Support

OfficeCLI ships as a self-contained binary with the .NET 10 runtime embedded — nothing to install at runtime. It is available for macOS (Apple Silicon and Intel), Linux (x64 and ARM64), and Windows (x64 and ARM64). Installation options include a one-line shell/PowerShell script, Homebrew (brew install officecli), npm (npm install -g @officecli/officecli), or manual binary download from GitHub Releases. Python and Node.js thin SDKs (officecli-sdk / @officecli/sdk) wrap the binary via resident named pipes for near-zero per-call latency.

Update: v1.0.129

The latest release is v1.0.129, published July 6, 2026, with 5,572 commits and 130 tags across 124 releases. Recent commits show active development across all three formats: Word mermaid diagram support (native editable shapes and full-fidelity PNG), Excel worksheet dump coverage for .xlsx, PowerPoint group ungroup, macOS notarization fixes for CoreCLR JIT on Hardened Runtime, and Node SDK quoting fixes for cmd.exe. The repository has 13 contributors and is written primarily in C# (94.4%).