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 PrithviSeran/hunch-mcp --skill document-editinggit clone --depth 1 https://github.com/PrithviSeran/hunch-mcpWrote 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/prithviseran/hunch-mcp/document-editing)<a href="https://agentmods.dev/skills/prithviseran/hunch-mcp/document-editing"><img src="https://agentmods.dev/badge/skills/prithviseran/hunch-mcp/document-editing/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/prithviseran/hunch-mcp/document-editing"><img src="https://agentmods.dev/badge/skills/prithviseran/hunch-mcp/document-editing.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 | $0.00028 | $0.01194 |
| Opus 5 | $0.00014 | $0.00597 |
| Sonnet 5 | $0.00006 | $0.00239 |
| Haiku 4.5 | $0.00003 | $0.00119 |
Grade A, and why
document-editing 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 6d ago.
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 — 88 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Edit documents in place
Use this workflow when editing documents in Google Docs, Microsoft Word, Apple Pages, or other native and web document editors. It is task-specific guidance, not an instruction to edit any document merely because its app is open. Preserve the user's document, tab, and editing mode. A request to revise or format a passage normally means editing that passage in the existing document, not creating a replacement document or copying the whole document into another chat. Whole-document rewriting is appropriate only when requested.
Locate and select
Choose the control layer appropriate to the actual app. Prefer declared document-range APIs or AppleScript operations when available; otherwise inspect native snapshot/act for native editors, and web_snapshot/web_act for browser editors. Read the live menus, outline, toolbar, and dialogs. Do not assume a Word, Pages, or other editor exposes Google's controls or shortcuts.
For browser editors, read web_snapshot for document tabs, outline headings, menus, toolbar controls, and dialogs. Click the relevant tab or heading by its current ref to navigate. Canvas text may be absent from the tree: use web_screenshot to see it and position the caret with web_act click_xy. On Safari, coordinates include browser chrome and must come from a fresh screenshot before every click or drag. Native act coordinates/keys are a different transport; use web_act for browser editing. Safari's window-routed input is not a generic native-app input backend. For native apps, use supported AX selection/actions or document APIs; if a needed operation requires gated foreground input, respect that restriction. Do not replace an entire native text field just to change a range.
For a distant passage, use the editor's Find command and enter a distinctive phrase from the requested passage. Inspect the match count and surrounding text; repeated phrases need more context. Close Find, then inspect the actual document selection/caret. A highlighted search match does not by itself prove the editor has a text selection. Do not type replacement text until the selection or insertion point is confirmed.
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.
- 6d ago First seen · 88 lines · 28 tokens per session scan A d802ae1303ec
document-editing is a skill published in the GitHub repository PrithviSeran/hunch-mcp (21 stars, last pushed 6d ago), licensed Apache-2.0. It adds 28 tokens to every session and 1,194 once invoked, about $0.0001 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 2026-09-14.
Other skills, from other repositories
macos-harness
Control a whole Mac from one persistent Python session with screenshots, PID-targeted input, an animated virtual pointer, targeted Apple Accessibility, Apple Events, Browser Harness CDP, and filesystem access. Use for native, Electron, browser, dialog, file, or cross-app tasks without moving the physical cursor or…
fetch-weather-report
Fetch the current weather forecast for a specified location and return a structured summary.
computer-use-macos
Top-level macOS computer-use skill with a bundled standalone runtime that bootstraps itself without any local Claude installation, private native modules, or extracted app assets.
unprocessed-ingest
A workflow for processing files from an `unprocessed/` folder, including screen recordings, exported conversations, notes, PDFs, and images. It moves information through raw and wiki stages.
control-window
Use when the user wants Claude to control, drive, or automate another app's window on macOS — clicking, typing, pressing keys, scrolling, or reading what's on screen in a window that isn't the terminal. Triggers on requests like "control my Telegram window", "drive this app for me", "click the button in that window"…
recipe-create-presentation
Create a new Google Slides presentation and add initial slides.