← Plugin catalog
Productivity

Letter.md

GUANXUN HUANG v1.0.0

Review Markdown with AI

Create a free account to save this plugin.

Competitor profile

Developer declarations; not independently tested. Listing last checked Nov 4, 2026. Package observed Nov 3, 2026.

Not specified in the available declarations: capabilities, limitations, pricing & account requirements.

Onboarding example

Review my Q3 Product Brief.

View package source →

Description

From the marketplace listing

Letter.md lets people read, search, and review Markdown documents from ChatGPT. It can create traceable comments and reviewable edit suggestions, inspect document images, compare versions, and open the complete realtime collaborative editor. With an explicit user request and publication permission, it can publish a document or update its public page; publishing never changes the private source document automatically. Users can also save generated Markdown as a new unpublished document in a writable Letter.md workspace or create a derivative document. Access is limited to the workspaces and scopes approved through OAuth.

Language: English · Automatically detected from descriptions.

What changed

Files & skills

File archives

Plugin package4 files · 2.66 KBBrowse files →
Skill instructions
letter-document-authoring2.17 KB

View saved version →

---
name: letter-document-authoring
description: Use when the user wants to create, find, open, read, or revise a Letter.md document. Guides safe document selection, creation, editor opening, and reviewable suggestions with Letter.md MCP tools.
---

# Letter.md document authoring

Use Letter.md as the source of truth for document identity and current content. Keep reads and writes separate so the user can understand what will change.

## Choose the document

- Use an explicit "documentId" when the user supplied one.
- Use "list_documents" when the title is ambiguous or the user wants to browse. Use "search_workspace_documents" when the user describes document content instead of its title.
- Use "render_letter_editor" without a document ID only when the user explicitly asks for their latest document.
- If multiple documents remain plausible, present the short candidates and ask the user to choose instead of guessing.

## Create a document

1. Draft complete Markdown before saving.
2. Call "create_document" only when the user asked to create or save a new Letter.md document. Supply a useful title and the finished Markdown. Omit "idempotencyKey" for normal retries.
3. Use the returned document ID with "render_letter_editor" so the user can continue in the collaborative editor.

Creating a new document never means overwriting an existing document. Do not use "create_document_from_ai" outside an active in-editor derivative workflow.

## Read or revise an existing document

1. Call "get_document" before quoting, summarizing, or proposing changes so the response uses the latest Markdown and block IDs.
2. For requested edits, call "create_suggestion" with narrow operations anchored to exact current block IDs and text. Prefer "replace_text" for a local edit and include prefix or suffix context when the same text can occur more than once.
3. Explain that a suggestion is reviewable and does not change the document until the user accepts it.
4. Call "render_letter_editor" when the user asks to open the document or when visual review and acceptance are the natural next step.

Do not claim that a suggestion was accepted or that document content changed unless a later tool result confirms it.
letter-document-review2.32 KB

View saved version →

---
name: letter-document-review
description: Use when the user wants to review, comment on, improve, or check publication readiness for a Letter.md document. Guides evidence-based review, comment reconciliation, and reviewable suggestions without silently editing the document.
---

# Letter.md document review

Review the current Letter.md document with evidence from its latest Markdown and block structure. Keep findings concise, non-overlapping, and actionable.

## Review workflow

1. Resolve the target document from an explicit ID, "list_documents", or "search_workspace_documents". Do not guess between similarly named documents.
2. Call "get_review_context" when comments are part of the request. Otherwise call "get_document" for the latest Markdown and block IDs.
3. Check structure, clarity, consistency, duplication, unsupported claims, and the user's stated goal. Quote only short exact excerpts and identify the relevant block IDs.
4. If the user requested changes, create one or more focused "create_suggestion" calls. Link comment-driven suggestions with "sourceCommentIds" and preserve exact anchors.
5. Open "render_letter_editor" when the user wants to inspect or accept the results.

## Publication readiness and publishing

- Use "get_publication_context" for title, slug, description, excerpt, indexing, current revisions, site state, and readiness recommendations.
- Treat publication preparation as read-only analysis. Never publish unless the user explicitly asks to make the document public or update its public version.
- For an unpublished document, call "publish_document" with the current document revision. Use a site handle only when the Workspace owner is creating the first public site.
- For an existing public version, call "update_document_publication" with both revisions returned by "get_publication_context". Never use the first-publish tool to overwrite a public version.
- If either tool reports that the document or public version changed, read the context again and show the user what changed before retrying.

## Safety and output

- Separate observations from proposed edits.
- Do not manufacture comments, citations, versions, or acceptance state.
- If no useful issue remains, say so instead of creating a cosmetic suggestion.
- Existing document edits remain reviewable suggestions until the user accepts them in Letter.md.
Package details

Publisher declarations from the archived package. These are separate from our research and the live service's terms.

Package author
GUANXUN HUANG

Package observed Nov 3, 2026.

Technical details
First seen
Oct 24, 2026 · 12:40 UTC
Last seen
Nov 4, 2026 · 06:00 UTC
Collection status
Collected

plugin_asdk_app_6a6029963b20819196ad13a9ca3a3a75

Download plugin data (JSON)