Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ilia-sokolov/OfficeAgent.NET --skill word-document-reviewgit clone --depth 1 https://github.com/ilia-sokolov/OfficeAgent.NETWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ilia-sokolov/officeagent.net/word-document-review)<a href="https://agentmods.dev/skills/ilia-sokolov/officeagent.net/word-document-review"><img src="https://agentmods.dev/badge/skills/ilia-sokolov/officeagent.net/word-document-review/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ilia-sokolov/officeagent.net/word-document-review"><img src="https://agentmods.dev/badge/skills/ilia-sokolov/officeagent.net/word-document-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00078 | $0.02876 |
| Opus 5 | $0.00039 | $0.01438 |
| Sonnet 5 | $0.00016 | $0.00575 |
| Haiku 4.5 | $0.00008 | $0.00288 |
Grade A, and why
word-document-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 218 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reviewing an existing Word document
OfficeAgent edits the real .docx package. A clause you replace keeps its numbering,
its comments, and its style; a redline you write is one Word shows in the review pane and
a human can accept or reject. That is the whole point of using it instead of extracting
text, rewriting it, and handing back a new file: the document that comes out is the one
that went in, minus the edit you made.
Working that way means never guessing where something is. Every edit names a location the server gave you and the text expected there, so an edit that would land in the wrong place fails instead.
First: how is the document reaching the server?
Look at what list_connections returns, or at the tools you have. This decides everything
that follows, and getting it wrong is the most common way a session falls apart.
| What you see | What it means | Use |
|---|---|---|
A connection with provider: "filesystem" or "sharepoint" |
The document is in storage the host controls | Address it by (connectionId, documentId). Edits save in place. |
A connection with provider: "session" |
The server holds documents in memory for this session | Same (connectionId, documentId) loop. import_document_content to put a document in, export_document_content once at the end. |
Only *_content tools, no connections |
Nothing is stored anywhere | create_document_content / edit_document_content, which carry the document as base64 both ways. |
Prefer a handle to the bytes
If both a session connection and the *_content tools are available, use the session
connection for anything beyond a single edit.
Chaining edit_document_content calls means passing the whole document back as base64 to
make the second edit — which requires reproducing several thousand characters exactly. That
does not survive contact with reality: a single wrong character makes the content stop
being a readable package, and every call after it fails on a copy you cannot repair,
because your copy is the damaged one. Retrying sends the same damaged string again.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday Changed · +3 lines 9e3aa51d8161
- 3d ago First seen · 215 lines · 78 tokens per session scan A 6b2b0e182c36
word-document-review is a skill published in the GitHub repository ilia-sokolov/OfficeAgent.NET (21 stars, last pushed 2d ago), licensed MIT. It adds 78 tokens to every session and 2,876 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-07.
Other skills, from other repositories
npoi
Use when reading/writing Excel (.xls/.xlsx) or Word (.doc/.docx) files in .NET without Microsoft Office — data import/export, report generation, template filling. NPOI: .NET port of Apache POI for Office file I/O in server environments.
aos-agent-skill-document
A document-publishing workflow for planning, creating, converting, checking, and preparing reports, proposals, manuals, white papers, and brand documents for release.
aos-author-word
Create, edit, inspect, convert, and visually verify professional Microsoft Word DOCX files while preserving editable structure. Use when Codex needs to author or locally revise reports, white papers, proposals, manuals, policies, brand books, tables, images, sections, headers, footers, high-fidelity image settings…
aos-process-pdf
Create, inspect, render, compare, clean, and visually verify professional PDF files while preserving fixed-layout fidelity. Use when Codex needs print-ready exports, PDF metadata and page-geometry inspection, page rendering, contact sheets, structural checks, public-release QA, or validation of PDF output converted…
aos-publish-document
Coordinate end-to-end professional document publishing across editable Word and fixed PDF deliverables. Use when Codex needs to turn source materials into publication-ready reports, white papers, proposals, manuals, brand books, or other business documents through content planning, DOCX authoring, PDF conversion…
opentelemetry-net-instrumentation
Provides guidance for implementing OpenTelemetry instrumentation in .NET codebases, covering tracing (Activities/Spans), metrics, logs, naming conventions, error handling, performance, SDK setup, resources, context propagation, and API design best practices.