Command Palette

Search for a command to run...

UnylyUnyly
Browse all

Dcc Mcp Word

FreeNot checked

Word adapter for the DCC-MCP ecosystem — headless Word IR to DOCX compile with read-back verification over the dcc-mcp-office runtime

GitHubEmbed

About

Word adapter for the DCC-MCP ecosystem — headless Word IR to DOCX compile with read-back verification over the dcc-mcp-office runtime

README

Word adapter for the DCC-MCP ecosystem — thin application layer over dcc-mcp-office.

Status: v0.1.0 — headless compile path live. Word IR → Open XML compile → read-back verification runs end to end with no Word installation, and the gate runs headless in CI. Field/TOC refresh, document reflow and track-changes inspection are host_limited: declared, graded, and never claimed as verified.

The old blocker note ("blocked on the dcc-mcp-office M1 release") is stale and removed. dcc-mcp-office shipped M1 in v0.2.0 (2026-08-16) and is now at v0.2.3; the Word IR (crates/office-ir, pub mod word, proposal §13.3) and the WordBackend.cs COM backend have been present since then. The three placeholder repos stalled on an unreviewed blocker note, not on a technical wall.

Part of the DCC-MCP host matrix

dcc-mcp-word — Word adapter for DCC-MCP — headless Word IR to DOCX compile with read-back verification over the dcc-mcp-office runtime.

It is one of 47 host adapters in the DCC-MCP catalog. Every adapter speaks the same MCP protocol and builds on the same core runtime contract; each one exposes the tools its own host needs on top of that.

This block is generated from the catalog entry in dcc-mcp-catalog.yml. Re-run the generator after changing the catalog.

What v0.1.0 does

Capability Grade Meaning
word.document.compile verified Word IR → DOCX, headless, CI-green
word.document.read_back verified Reopen the artifact; compare every paragraph and table cell
word.paragraphs.write, word.lists.write, word.tables.write verified Structured content writes
word.headers_footers.write, word.fields.write verified Per-section headers/footers; real Word field instructions
word.fields.update, word.toc.rebuild host_limited Word computes field results — needs the desktop app
word.document.reflow, word.track_changes.inspect host_limited Needs the Word COM object model
word.content_controls.fill, word.figures.inline unimplemented Parsed and validated by the IR, not emitted by the writer

Run dcc-mcp-word capabilities for the machine-readable report, or dcc-mcp-word preflight for the per-machine self-check.

Install

pip install dcc-mcp-word          # adapter only (stdlib + dcc-mcp-core)
pip install "dcc-mcp-word[headless]"   # + python-docx, for the compile path

python-docx is an opt-in extra, matching how dcc-mcp-excel treats openpyxl: importing the package never pulls it, so the adapter stays installable inside DCCs that pin their own version. A test enforces this (tests/test_runtime_purity.py).

Quick start

# 1. a Word IR envelope (office-ir/1.0, kind: word)
cat > call-sheet.json <<'JSON'
{
  "schema_version": "office-ir/1.0",
  "kind": "word",
  "document_id": "fpt:call-sheet-2026-10-07",
  "metadata": {"title": "Call sheet — 2026-10-07", "author": "pipeline"},
  "document": {
    "sections": [{"title": "Schedule", "paragraphs": [{"text": "Call 07:00"}]}],
    "tables": [{"header": true, "rows": [["Shot", "Status"], ["sh010", "ip"]]}],
    "footers": [{"section_index": 0, "text": "Page"}],
    "fields": [{"kind": "toc"}, {"kind": "page"}]
  }
}
JSON

# 2. compile + read back
python -m dcc_mcp_word.skills.word-document.scripts.generate_document \
  --input call-sheet.json --out output

The script prints JSON: artifact path, read-back verdict (paragraphs and cells checked, mismatches), validation checks and warnings.

Input contract

The adapter mirrors the Rust dcc-mcp-office-ir word schema 1:1 — styles, sections, paragraphs, lists, tables, figures, content controls, headers/footers, fields and review policy. See src/dcc_mcp_word/word_ir.py for the annotated contract, or word-document/SKILL.md for the agent-facing guide.

The honest boundary

python-docx writes a valid DOCX but has no layout engine. That is why the two halves of word.document.reflow are graded separately:

  • Compiling (IR → DOCX) is headless and verified.
  • Reflowing, re-paginating, and refreshing field results needs Word, and is host_limited.

A TOC field is written as a real field instruction. Word fills in the entries when the document is opened — verified on a Word 365 host: before Fields.Update() the TOC shows its placeholder, after it shows Schedule→1, Safety→2. See examples/com-verification.json.

Test

pip install -e ".[dev]"
pytest
ruff check src tests

The headless gate must stay green without Word installed — that is the proof the adapter does not depend on a desktop Word installation.

Upstream

  • dcc-mcp-office — protocol, IR, C# COM runtime, jobs, security policy
  • dcc-mcp-core — gateway, skills runtime, sidecar lifecycle

License

MIT

from github.com/dcc-mcp/dcc-mcp-word

Install Dcc Mcp Word in Claude Desktop, Claude Code & Cursor

Recommended · one command, every IDE
unyly install dcc-mcp-word

Installs into Claude Desktop, Claude Code, Cursor & VS Code — handles npx, uvx and build-from-source repos for you.

First time? Get the CLI: curl -fsSL https://unyly.org/install | sh

Or configure manually

Run in your terminal:

claude mcp add dcc-mcp-word -- uvx dcc-mcp-word

Step-by-step: how to install Dcc Mcp Word

FAQ

Is Dcc Mcp Word MCP free?

Yes, Dcc Mcp Word MCP is free — one-click install via Unyly at no cost.

Does Dcc Mcp Word need an API key?

No, Dcc Mcp Word runs without API keys or environment variables.

Is Dcc Mcp Word hosted or self-hosted?

Self-hosted: the server runs locally on your machine via the install command above.

How do I install Dcc Mcp Word in Claude Desktop, Claude Code or Cursor?

Open Dcc Mcp Word on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.

Related MCPs

Compare Dcc Mcp Word with

Not sure what to pick?

Find your stack in 60 seconds

Author?

Embed badge for your README

Browse similar

All ai MCPs