- Home
- Skills
- Workflow Automation
- Google Chat Workspace Agent (ADK + Workspace MCP)
Works with the AI Agents you already use
Google Chat Workspace Agent (ADK + Workspace MCP)
Build an agent in Google Chat that reads mail, schedules meetings and searches Drive as the person asking: request verification, async replies, per-user OAuth, Workspace MCP and safe deployment on Cloud Run.
$15
Google Chat Workspace Agent (ADK + Workspace MCP)
Example session with this skill installed
We want a Google Chat bot for our 200-person Google Workspace company. People should be able to DM it things like "what's on my calendar tomorrow and draft a reply to the last email from finance". It should run on Cloud Run with Gemini. Write the design and security decisions in DESIGN.md and the OAuth scopes in scopes.txt.
- Read your context and instructions
- Compiled the google chat workspace
DESIGN.md: webhook verifies Chat's bearer token (audience = endpoint URL, classic or add-on issuer), returns 200 immediately and posts the answer with spaces.messages.create from a background task; first-time users get an auth card; consent uses signed 15-minute state, offline access and stored refresh tokens keyed by Chat user id; one ADK agent with Gmail and Calendar MCP toolsets whose header provider resolves the caller's token per call; drafting and sending email or creating events requires confirmation; min-instances 1 and CPU always allocated. scopes.txt: openid, userinfo.email, cloud-platform, gmail.readonly, gmail.compose, gmail.modify, calendar.calendarlist.readonly, calendar.events.freebusy, calendar.events.
Connects securely to your tools. The creator never sees your data.
About this skill
An agent inside Google Chat is only useful when it can act as the person asking: their inbox, their calendar, their Drive. That requirement drives the whole design: Chat events carry no user token, Chat gives you about 30 seconds to answer, and one deployment serves the whole domain. This skill gets those three right.
What your agent builds with it
- Request verification for both classic Chat apps and Workspace add-ons (two different token issuers), failing closed.
- Asynchronous replies: acknowledge instantly, run the agent in the background, post the answer via the Chat API, so long tool-calling turns never time out.
- Per-user OAuth: signed, expiring
statebound to the Chat user, offline consent, refresh tokens stored per user, scope-drift detection that triggers re-consent. - One shared ADK agent whose Workspace MCP toolsets resolve the caller's token on every tool call, so users never share sessions.
- Scope planning with the non-obvious lessons encoded (e.g. why
cloud-platformis needed, why read-only scopes next to write tools fail). - User-connected MCP servers (paste a URL or a Claude/VS Code config into the chat) with stdio refused, https-only and an optional host allow-list.
- Deployment: Cloud Run flags that matter, Chat API configuration, slash commands, local tunnel testing and a smoke test.
Included: SKILL.md, code patterns and setup references, mcp_spec.py (pasted-config parser), oauth_state.py (signed state), scope_plan.py (scope planner and re-consent check), and an eval case. Scripts are standard-library Python.
Part of the Production AI Agents series. Pair with MCP Server in Production and Agent Guardrails; start free with Agent Architecture.
Google Chat only (not Slack or Teams). Workspace MCP availability and scope names change; the skill dates its scope table and tells you to verify against Google's docs. Does not cover Marketplace publishing.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
Requires
Permissions
File Scopes
Bundled scripts run locally with the Python standard library and make no network calls.