Documentation Index
Fetch the complete documentation index at: https://docs.pinksheep.ai/llms.txt
Use this file to discover all available pages before exploring further.
Prerequisites
- A Pinksheep account (sign up here)
- At least one business tool you want to automate (e.g. HubSpot, Gmail, Slack, Google Sheets)
1. Select your stack
After signing up, you’ll be asked to select the apps your business uses. This is your stack. Pinksheep uses it to scope which tools are available to your agents and to recommend relevant templates.2. Start from a template or build from scratch
From a template: Browse the Template Library and pick an agent that matches your use case. Templates come pre-configured with steps, permissions, and eval scenarios. Click “Use this template” to start. From scratch: Go to the Builder and describe what you want. Use @mentions to reference specific apps:“Every Monday at 9am, pull all open deals from @HubSpot, flag any with no activity in 7 days, and post a summary to #sales in @Slack.”
3. Connect your apps
The builder detects which apps your agent needs. For each one, click Connect and complete the OAuth flow. Pinksheep verifies each connection server-side before proceeding.4. Review the agent plan
Once the builder has enough context, it produces a manifest showing:| Field | What it means |
|---|---|
| Steps | The ordered actions your agent will take |
| Permissions | What data the agent reads, writes, and whether any actions are irreversible |
| Schedule | When the agent runs (if scheduled) |
| Estimated credits | Cost range per execution |
| Monthly projection | Projected monthly spend based on frequency and credit cost |
5. Test the agent
Click Test this agent to run a sandboxed dry-run. The test executes against your real connections but in a safe context. You’ll see:- Whether each step succeeded or failed
- A result summary
- The credit cost of the test
6. Deploy
Once the test passes, click Approve & Deploy. Your agent is now live. You can find it on your Dashboard. Scheduled agents will run automatically. Manual agents can be triggered from the dashboard or via the API.What happens next
- Run notifications: You’ll be notified when runs complete, fail, or need approval.
- Monitoring: View run history, success rates, and credit usage on the agent detail page.
- Fixing failures: If an agent fails, use the “Fix this agent” button to re-enter the builder with full failure context pre-loaded.
Next steps
Concepts
Learn about autonomy levels, credits, and how execution works.
Stack Selection
Configure which tools are available to your agents.