composio-google-docs
GitHub提供通过 Composio 操作 Google Docs 的指导,区分纯文本与结构 JSON 读取工具,说明参数配置及文档 ID 获取方法。
Trigger Scenarios
Install
npx skills add memoe-app/memoe --skill composio-google-docs -g -y
SKILL.md
Frontmatter
{
"name": "composio-google-docs",
"author": "Moe",
"license": "MIT",
"version": "1.0.0",
"metadata": {
"hermes": {
"tags": [
"Composio",
"Google Docs",
"Documents",
"MCP"
],
"toolkit": "googledocs",
"related_skills": [
"composio-router",
"composio-google-drive"
]
}
},
"description": "Read before any GOOGLEDOCS_* call through Composio. There are two tools for 'read this document' and they answer different questions: one returns plain text, the other returns the document's whole structural JSON. The preloaded one is the structural one."
}
Google Docs through Composio
Read composio-router first for the method.
Byte costs on this page are unverified. The schema facts below were read from Composio on 2026-09-04 and are properties of the tools, true for anyone. Sizes are not, because measuring them needs a connected Google Docs account and this machine had none. Where a size matters, take it yourself: ask for one document, look at the result, then decide.
The two read tools, and why the wrong one is the obvious one
| the question | the tool |
|---|---|
| what does this document say | GOOGLEDOCS_GET_DOCUMENT_PLAINTEXT |
| where is the heading / what is the structure / who commented | GOOGLEDOCS_GET_DOCUMENT_BY_ID |
GOOGLEDOCS_GET_DOCUMENT_BY_ID is the one preloaded into the tool list, so
it is the one you will reach for without thinking. It returns the Google Docs
API document resource: a nested JSON tree of structural elements, paragraphs,
text runs and style objects. Every sentence of prose arrives wrapped in
several layers of it.
GOOGLEDOCS_GET_DOCUMENT_PLAINTEXT is not preloaded — reach it through
COMPOSIO_MULTI_EXECUTE_TOOL — and it returns a best-effort plain-text
rendering of the same document. If the person asked what a document says,
summarise it, or find something in it, this is the tool.
Its switches, and what they default to
GOOGLEDOCS_GET_DOCUMENT_PLAINTEXT:
include_tables default TRUE tables rendered with delimiters
include_headers default false
include_footers default false
include_footnotes default false
include_tabs_content default false main body only
table_cell_delimiter default "\t"
table_row_delimiter default "\n"
These are content switches, not size switches, and their defaults are sensible: you get the body and its tables, and none of the running headers. Turn one on when the person's question is about that part.
GOOGLEDOCS_GET_DOCUMENT_BY_ID has exactly one switch,
includeTabsContent (default false). With it false the API returns the
legacy first-tab fields; with it true you get every tab. A multi-tab document
read at the default silently gives you only the first tab — which reads as a
short document rather than as an error.
Both accept a document id or a full Docs URL.
Finding the document
Docs tools need an id. Get it from GOOGLEDRIVE_FIND_FILE with a
mimeType = 'application/vnd.google-apps.document' filter, or from a URL the
person gave you. Do not download a Doc through Drive — a Google Doc has no
file bytes; export it, or use these tools.
Version History
- e2735d6 Current 2026-11-20 21:42


