home / skills / openclaw / skills / scalekit-agent-auth
This skill manages secure OAuth tokens via Scalekit, refreshing automatically and returning fresh tokens for Gmail, Slack, GitHub, and other services.
npx playbooks add skill openclaw/skills --skill scalekit-agent-authReview the files below or copy the command above to add this skill to your agents.
---
name: scalekit-auth
description: Secure OAuth token management via Scalekit. Handles token storage, refresh, and retrieval for third-party services (Gmail, Slack, GitHub, etc.). Never stores tokens locally - always fetches fresh tokens from Scalekit.
homepage: https://scalekit.com
metadata:
openclaw:
requires:
bins: ["python3"]
install:
- id: python-deps
kind: exec
command: "pip3 install scalekit-sdk-python python-dotenv requests"
label: "Install Python dependencies"
---
# Scalekit Auth - Secure Token Management
Centralized OAuth token management for AI agents. No local token storage, automatic refresh, multi-service support.
## Why Use This?
**Problem:** OAuth tokens scattered across config files, no refresh logic, security risks.
**Solution:** Scalekit handles all token lifecycle:
- ✅ Secure cloud storage (never stored locally)
- ✅ Automatic token refresh
- ✅ Multi-service support (Gmail, Slack, Notion, GitHub, etc.)
- ✅ Always returns fresh, valid tokens
## Installation
### 1. Install Skill
```bash
clawhub install scalekit-auth
cd skills/scalekit-auth
pip3 install -r requirements.txt
```
### 2. Get Scalekit Credentials
1. Sign up at [scalekit.com](https://scalekit.com)
2. Go to Dashboard → Developers → Settings → API Credentials
3. Copy:
- Client ID
- Client Secret
- Environment URL
### 3. Configure Credentials
Create `skills/scalekit-auth/.env`:
```bash
SCALEKIT_CLIENT_ID=your_client_id_here
SCALEKIT_CLIENT_SECRET=your_client_secret_here
SCALEKIT_ENV_URL=https://your-env.scalekit.com
```
**Or** let the agent ask you on first use.
## Setting Up a Service (e.g., Gmail)
### Step 1: Create Connection in Scalekit Dashboard
1. Go to Scalekit Dashboard → Connections → Add Connection
2. Select provider (e.g., Gmail/Google)
3. Configure OAuth:
- Get Client ID/Secret from Google Cloud Console
- Set Redirect URI (provided by Scalekit)
4. **Copy the `connection_name`** (e.g., `gmail_u3134a`)
### Step 2: Register with Agent
Tell the agent:
```
"Configure Gmail for Scalekit. Connection name is gmail_u3134a"
```
Agent stores it in `connections.json`:
```json
{
"gmail": {
"connection_name": "gmail_u3134a",
"identifier": "mess"
}
}
```
### Step 3: Authorize
First API call will prompt:
```
Authorization needed for Gmail.
Link: https://scalekit.com/auth/... (expires in 1 minute!)
```
Click link → authorize → done!
## Usage
### From Agent Skills
```python
#!/usr/bin/env python3
import sys
sys.path.append('./skills/scalekit-auth')
from scalekit_helper import get_token
# Get fresh token for any service
access_token = get_token("gmail")
# Use it immediately
headers = {"Authorization": f"Bearer {access_token}"}
response = requests.get("https://gmail.googleapis.com/gmail/v1/users/me/messages", headers=headers)
```
### From Shell Scripts
```bash
# Get token via CLI wrapper
TOKEN=$(python3 skills/scalekit-auth/get_token.py gmail)
# Use in API call
curl -H "Authorization: Bearer $TOKEN" \
https://gmail.googleapis.com/gmail/v1/users/me/messages
```
## Configuration Files
### connections.json
Maps service names to Scalekit connection names:
```json
{
"gmail": {
"connection_name": "gmail_u3134a",
"identifier": "mess"
},
"slack": {
"connection_name": "slack_x7y9z",
"identifier": "mess"
}
}
```
**Note:** `identifier` is auto-set to agent's name (from IDENTITY.md).
### .env
Scalekit API credentials (never commit to git!):
```bash
SCALEKIT_CLIENT_ID=sk_live_...
SCALEKIT_CLIENT_SECRET=...
SCALEKIT_ENV_URL=https://...
```
## Supported Services
Any OAuth provider Scalekit supports:
- Gmail, Google Calendar, Google Drive
- Slack, Notion, Linear, GitHub
- Salesforce, HubSpot, Zendesk
- 50+ more
Check [Scalekit Connectors](https://docs.scalekit.com/connectors) for full list.
## Authorization Flow
```
1. Agent calls get_token("gmail")
2. Check if connection configured → if NO, ask user
3. Check if authorized (status == ACTIVE)
4. If NOT authorized:
- Generate auth link (expires 1 min)
- Send to user via Telegram/chat
- Wait for authorization
5. Return fresh access_token
6. Scalekit auto-refreshes in background
```
## Error Handling
**Connection not configured:**
```
Error: gmail not configured. Please:
1. Create connection in Scalekit dashboard
2. Provide connection_name
```
**Authorization expired:**
```
Authorization needed: [link]
(Link expires in 1 minute - click now!)
```
**Scalekit credentials missing:**
```
Scalekit not configured. Please provide:
- SCALEKIT_CLIENT_ID
- SCALEKIT_CLIENT_SECRET
- SCALEKIT_ENV_URL
```
## Security Best Practices
1. **Never log tokens** - use `[REDACTED]` in logs
2. **Add .env to .gitignore** - never commit credentials
3. **Rotate credentials** if exposed
4. **Use separate Scalekit accounts** for dev/prod
5. **Auth links expire in 1 min** - act fast!
## Troubleshooting
**"Module not found" error:**
```bash
cd skills/scalekit-auth
pip3 install -r requirements.txt
```
**Token returns 401:**
- Authorization may have expired
- Agent will prompt for re-authorization
**Connection not found:**
- Check `connections.json` exists
- Verify connection_name from Scalekit dashboard
## Example: Gmail Integration
```python
# In your skill's script
from scalekit_helper import get_token
import requests
def fetch_unread_emails():
token = get_token("gmail")
headers = {"Authorization": f"Bearer {token}"}
url = "https://gmail.googleapis.com/gmail/v1/users/me/messages"
params = {"q": "is:unread", "maxResults": 5}
response = requests.get(url, headers=headers, params=params)
return response.json()
```
## Publishing Skills with Scalekit Auth
If your skill uses scalekit-auth:
1. **Document in SKILL.md:**
```markdown
## Prerequisites
- Install scalekit-auth skill
- Configure [SERVICE] connection in Scalekit
```
2. **Import in scripts:**
```python
sys.path.append('./skills/scalekit-auth')
from scalekit_helper import get_token
```
3. **Handle errors gracefully** - guide users to configure connections
## API Reference
### get_token(service_name: str) → str
Returns fresh OAuth access token for the service.
**Parameters:**
- `service_name`: Service identifier (e.g., "gmail", "slack")
**Returns:**
- `access_token`: Fresh OAuth bearer token
**Raises:**
- `ConfigurationError`: Service not configured or Scalekit creds missing
- `AuthorizationError`: User needs to authorize (sends link to user)
**Example:**
```python
token = get_token("gmail")
print(f"Token: {token[:10]}...") # Never log full token!
```
## Roadmap
- [ ] Multi-user support (multiple identifiers per service)
- [ ] Token caching (reduce API calls)
- [ ] CLI tool (`scalekit-auth config gmail gmail_u3134a`)
- [ ] Auto-detect service from API URL
- [ ] Batch token retrieval
## Contributing
Found a bug? Have a feature request? Open an issue on ClawHub!
---
**Remember:** Tokens are secrets. Handle with care. 🔐
This skill provides centralized OAuth token management via Scalekit for AI agents. It never stores tokens locally, always fetches fresh access tokens, and handles automatic refresh and multi-service support. Use it to simplify secure access to Gmail, Slack, GitHub, and other OAuth providers.
The skill talks to your Scalekit environment using configured API credentials and a connections mapping. When an agent calls get_token(service_name), the skill checks connection configuration, verifies authorization, and either returns a fresh access token or generates a short-lived auth link for user approval. Scalekit performs background refreshes so returned tokens are valid on use.
What happens if a service is not configured?
The skill raises a configuration error and instructs you to create a connection in the Scalekit dashboard and add the connection_name to connections.json.
How are authorization links delivered?
When authorization is required, the skill generates a short-lived link and sends it to the user through the configured agent channel; the link typically expires in about one minute.
Are tokens stored locally?
No. Tokens are never stored locally; the skill always requests fresh tokens from Scalekit and relies on Scalekit for secure storage and automatic refresh.