Remote OpenClaw
Menu
SkillsMCPPluginsGuideAdvertise
Remote OpenClaw
SkillsMCPPluginsGuide Advertise

Featured

Launch your own AI agent in one click logoLaunch your own AI agent in one click

A live, always-on OpenClaw agent that handles your tasks around the clock. No servers, no setup — pick a model, connect Telegram, and it starts working.

Try it for free →
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you →
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free →
CLN.Work — Stop prompting, start hiring AI employees logoCLN.Work — Stop prompting, start hiring AI employees

Turn your Claude agents into a real team — onboard them, assign tasks, and manage them like staff.

Hire AI employees →
Keep your OpenClaw agent online 24/7 logoKeep your OpenClaw agent online 24/7

Your OpenClaw or Hermes agent, live around the clock on a Hostinger VPS. Set up in about 60 seconds, from a few dollars a month, with 20% off through this link.

Get hosting →
Launch your AI product and start charging today logoLaunch your AI product and start charging today

Auth, billing, and AI already wired in. Skip months of boilerplate and get paying customers now, not next quarter.

See the kit →
Launch your own AI agent in one click logoLaunch your own AI agent in one click

A live, always-on OpenClaw agent that handles your tasks around the clock. No servers, no setup — pick a model, connect Telegram, and it starts working.

Try it for free →
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you →
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free →
CLN.Work — Stop prompting, start hiring AI employees logoCLN.Work — Stop prompting, start hiring AI employees

Turn your Claude agents into a real team — onboard them, assign tasks, and manage them like staff.

Hire AI employees →
Keep your OpenClaw agent online 24/7 logoKeep your OpenClaw agent online 24/7

Your OpenClaw or Hermes agent, live around the clock on a Hostinger VPS. Set up in about 60 seconds, from a few dollars a month, with 20% off through this link.

Get hosting →
Launch your AI product and start charging today logoLaunch your AI product and start charging today

Auth, billing, and AI already wired in. Skip months of boilerplate and get paying customers now, not next quarter.

See the kit →
Launch your own AI agent in one click logoLaunch your own AI agent in one click

A live, always-on OpenClaw agent that handles your tasks around the clock. No servers, no setup — pick a model, connect Telegram, and it starts working.

Try it for free →
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you →
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free →
CLN.Work — Stop prompting, start hiring AI employees logoCLN.Work — Stop prompting, start hiring AI employees

Turn your Claude agents into a real team — onboard them, assign tasks, and manage them like staff.

Hire AI employees →
Keep your OpenClaw agent online 24/7 logoKeep your OpenClaw agent online 24/7

Your OpenClaw or Hermes agent, live around the clock on a Hostinger VPS. Set up in about 60 seconds, from a few dollars a month, with 20% off through this link.

Get hosting →
Launch your AI product and start charging today logoLaunch your AI product and start charging today

Auth, billing, and AI already wired in. Skip months of boilerplate and get paying customers now, not next quarter.

See the kit →
Skills/caffeinelabs/skills/connector-googlemail

connector-googlemail

caffeinelabs/skills
668 installs0 stars

Installation

npx skills add https://github.com/caffeinelabs/skills --skill connector-googlemail

Summary

>-

SKILL.md

googlemail-client

Motoko bindings for the Gmail API v1, generated from Google's official OpenAPI spec.

Send-focused PoC surface: gmail_users_messages_send, gmail_users_drafts_{create,send,get,list}, gmail_users_messages_{get,list}, gmail_users_getProfile. All 8 operations live in Apis/UsersApi.mo.

Trigger phrases

Reach for this skill on any request mentioning: send email, send message, compose email, Gmail, draft, inbox, mailbox, "email the user", "notify via email", "forward results by email", "send a notification email".

How Gmail authentication works (read before wiring)

Gmail uses OAuth 2.0 Authorization Code flow — there is no static API key. The canister never mints a token on its own; the user completes an OAuth dance off-chain and passes in the resulting Bearer access token at call time.

Token lifetime: 1 hour by default (Google's access tokens). After expiry the API returns HTTP 401. The refresh token must be exchanged off-chain too — the canister cannot call Google's token endpoint (that would expose the client secret on-chain). Surface a #Err("auth_expired") result and ask the caller to re-authenticate.

Required OAuth 2.0 scope for messages.send: https://www.googleapis.com/auth/gmail.send. For read access add: https://www.googleapis.com/auth/gmail.readonly.

Usage

import { gmail_users_messages_send; gmail_users_getProfile;
         gmail_users_drafts_create; gmail_users_drafts_send }
  "mo:googlemail-client/Apis/UsersApi";
import { Message; type Message } "mo:googlemail-client/Models/Message";
import { Draft; type Draft } "mo:googlemail-client/Models/Draft";
import { defaultConfig } "mo:googlemail-client/Config";
import Text "mo:core/Text";  // Text.encodeUtf8: build the raw RFC 2822 Blob

// Shared cfg — swap in the caller's short-lived bearer token.
let cfg = {
  defaultConfig with
    auth               = ?#bearer "<off-chain OAuth2 access token>";
    max_response_bytes = ?500_000;
    is_replicated      = ?false; // non-replicated: required for sends (see Notes); reads too
};

// Send a message. `raw` is the PLAIN RFC 2822 message as a Blob — the client
// base64-encodes it for the Gmail API; do NOT base64-encode it yourself.
let mime : Text = "From: me\r\nTo: friend@example.com\r\nSubject: Hi\r\n\r\nHello!";
let outMsg = Message.init {};   // all-null base, then layer fields:
let envelope : Message = { outMsg with raw = ?Text.encodeUtf8(mime) };
let result = await* gmail_users_messages_send(cfg,
  "me",         // userId: "me" = authenticated user
  #_1_,         // $.xgafv — use #_1_ (v1) for all calls
  "",           // accessToken — leave "" when auth = ?#bearer above
  #json,        // alt
  "", "", "", "", true, "", "", "",   // callback/fields/key/oauthToken/prettyPrint/quotaUser/uploadProtocol/uploadType
  envelope
);

// Get the authenticated user's email address
let profile = await* gmail_users_getProfile(cfg, "me",
  #_1_, "", #json, "", "", "", "", true, "", "", "");
let ?email = profile.emailAddress else return #Err("no email");

Notes

  • Use is_replicated = ?false (non-replicated) for sends (gmail_users_messages_send,

gmail_users_drafts_send). These outcalls are non-idempotent and Gmail's response is non-deterministic (unique message id, per-request Date header). In replicated mode (null) every subnet replica issues the request — so the email is sent once per replica (duplicates) and the differing responses fail IC consensus ("No consensus could be reached. Replicas had different responses"). Non-replicated has a single node perform exactly one send. Reads (gmail_users_messages_list, gmail_users_messages_get, gmail_users_getProfile) also use ?false (one node, ~13× cheaper).

  • The $.xgafv parameter (version discriminator) should always be #_1_ for

Gmail API v1 calls. The alt parameter should always be #json.

  • All optional string parameters (callback, fields, key, oauthToken,

quotaUser, uploadProtocol, uploadType) accept "" to omit them; prettyPrint can be false.

  • userId = "me" refers to the authenticated user. Explicit email addresses

also work but require the https://mail.google.com/ scope.

  • Messages must be in RFC 2822 format, passed as a plain Blob in the raw

field (e.g. ?Text.encodeUtf8(mime)). The client base64-encodes raw for the API — do not base64-encode it yourself (that double-encodes and Gmail rejects it). The Message.payload / MessagePart fields are for parsed read responses — don't try to build them for sending.

  • Google returns HTTP 429 on rate-limit (quota exceeded). Surface the error to

the caller; never silently retry inside the canister — a send retry may deliver duplicates.

  • Access tokens expire in 1 hour. On 401, surface #Err("auth_expired") so

the caller can re-authenticate off-chain and retry with a fresh token.

  • Draft.message holds a Message; gmail_users_drafts_create builds a draft

server-side. Use gmail_users_drafts_send to send a draft by its id.

  • max_response_bytes: Gmail message reads can be large. 500 KB covers typical

messages; bump to 2 MB for messages with large payloads (Gmail API's max response is bounded by the API, but set conservatively for cycle budgets).

  • Cycle budget: defaultConfig.cycles = 30_000_000_000 (30B). On the IC,

outbound HTTPS calls cost ~10–15B cycles for a typical send. Adjust if you see InsufficientCycles errors.

Featured

Launch your own AI agent in one click logoLaunch your own AI agent in one click

A live, always-on OpenClaw agent that handles your tasks around the clock. No servers, no setup — pick a model, connect Telegram, and it starts working.

Try it for free →
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you →
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free →
CLN.Work — Stop prompting, start hiring AI employees logoCLN.Work — Stop prompting, start hiring AI employees

Turn your Claude agents into a real team — onboard them, assign tasks, and manage them like staff.

Hire AI employees →
Keep your OpenClaw agent online 24/7 logoKeep your OpenClaw agent online 24/7

Your OpenClaw or Hermes agent, live around the clock on a Hostinger VPS. Set up in about 60 seconds, from a few dollars a month, with 20% off through this link.

Get hosting →
Launch your AI product and start charging today logoLaunch your AI product and start charging today

Auth, billing, and AI already wired in. Skip months of boilerplate and get paying customers now, not next quarter.

See the kit →
Launch your own AI agent in one click logoLaunch your own AI agent in one click

A live, always-on OpenClaw agent that handles your tasks around the clock. No servers, no setup — pick a model, connect Telegram, and it starts working.

Try it for free →
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you →
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free →
CLN.Work — Stop prompting, start hiring AI employees logoCLN.Work — Stop prompting, start hiring AI employees

Turn your Claude agents into a real team — onboard them, assign tasks, and manage them like staff.

Hire AI employees →
Keep your OpenClaw agent online 24/7 logoKeep your OpenClaw agent online 24/7

Your OpenClaw or Hermes agent, live around the clock on a Hostinger VPS. Set up in about 60 seconds, from a few dollars a month, with 20% off through this link.

Get hosting →
Launch your AI product and start charging today logoLaunch your AI product and start charging today

Auth, billing, and AI already wired in. Skip months of boilerplate and get paying customers now, not next quarter.

See the kit →
Launch your own AI agent in one click logoLaunch your own AI agent in one click

A live, always-on OpenClaw agent that handles your tasks around the clock. No servers, no setup — pick a model, connect Telegram, and it starts working.

Try it for free →
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you →
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free →
CLN.Work — Stop prompting, start hiring AI employees logoCLN.Work — Stop prompting, start hiring AI employees

Turn your Claude agents into a real team — onboard them, assign tasks, and manage them like staff.

Hire AI employees →
Keep your OpenClaw agent online 24/7 logoKeep your OpenClaw agent online 24/7

Your OpenClaw or Hermes agent, live around the clock on a Hostinger VPS. Set up in about 60 seconds, from a few dollars a month, with 20% off through this link.

Get hosting →
Launch your AI product and start charging today logoLaunch your AI product and start charging today

Auth, billing, and AI already wired in. Skip months of boilerplate and get paying customers now, not next quarter.

See the kit →

Categories

External DownloadsPrompt Injection
View on GitHub

Recommended skills

Browse all →

find-skills

vercel-labs/skills

2.3M installsInstall

frontend-design

anthropics/skills

625K installsInstall

vercel-react-best-practices

vercel-labs/agent-skills

525K installsInstall

agent-browser

vercel-labs/agent-browser

511K installsInstall

grill-me

mattpocock/skills

453K installsInstall

web-design-guidelines

vercel-labs/agent-skills

438K installsInstall

Browse

Skills by category

Frontend250Git198Data154Testing120Design105Docs103Security96Automation87Backend76Devops37Productivity29Mcp23

Advertise on Remote OpenClaw

Get your AI tool in front of 67,000+ AI enthusiasts a month

See placements & pricing →

Remote OpenClaw

AI agent skills directory, marketplace, and workflow hub for OpenClaw, Hermes Agent, Claude Code, Codex, and MCP-powered operator stacks.

Explore

  • Home
  • Skills Directory
  • Claude Code Skills
  • Codex Skills
  • Marketplace
  • Hermes Ecosystem
  • Guide
  • Learn
  • Blog

More

  • Playbook
  • Free Tools
  • Shipping
  • Contact
  • Terms
  • Privacy
© 2026 Remote OpenClaw
Fazier badgeFeatured on Twelve ToolsFeatured on Wired BusinessRemote OpenClaw - Featured on AI Agents DirectoryListed on Turbo0