AI Playbooks
Two senses of "playbook" — don't confuse them. This page describes conceptual flows (narratives for how an AI operates the platform). The instantiable playbook templates an MCP client fills in and runs (
find_playbooks/instantiate_playbook) are a separate, concrete set — see MCP Playbooks → Available templates (CONTRACT_ANCHOR,UI_FLOW,PERF_FROM_API,DATA_DRIVEN). The pages here are the why; that page is the what you can actually instantiate.
Proofarc is API-first. Every action in the platform — creating projects, running tests, scoring releases, generating compliance reports — is a REST API call. This means AI assistants like Claude, ChatGPT, or Copilot can fully operate the platform.
How It Works
- Give your AI assistant the Proofarc API endpoint and credentials
- Describe what you want in natural language
- The AI translates your request into API calls
- Results come back as structured JSON
Example
You say:
"Create an API test for the user-service at http://user-service:8089. Test the full CRUD lifecycle — create a user, read it back, update the email, verify the update, then delete. Run it and tell me the results."
The AI does:
# 1. Authenticate
POST /api/auth/login → token
# 2. Create scenario with 5 steps
POST /api/scenarios → scenarioId
# 3. Execute
POST /api/scenarios/{id}/execute → executionId
# 4. Get results
GET /api/scenarios/executions/{id} → 5/5 passed
You get: "All 5 steps passed. User created, read back with correct email, updated successfully, update verified, deleted. Full CRUD lifecycle working."
Available Playbooks
| Playbook | What it does |
|---|---|
| Onboard a Project | Create project, environments, targets, first test |
| API Test from Swagger | Parse Swagger spec, generate CRUD scenarios |
| Full Release Validation | Run all tests, score readiness, check compliance |
| Investigate Failures | Analyze failed tests, check logs, find root cause |
API Base URL
Production: https://app.proofarc.ai
Local: http://localhost:8080
Authentication
Every API call needs a Bearer token:
# Get token
TOKEN=$(curl -s $BASE_URL/api/auth/login -X POST \
-H "Content-Type: application/json" \
-d '{"username":"admin","password":"your-password"}' | \
python3 -c "import sys,json;print(json.load(sys.stdin)['token'])")
# Use token
curl -H "Authorization: Bearer $TOKEN" $BASE_URL/api/projects
Why API-First Matters
Most testing platforms are UI-first — you click buttons, fill forms, run tests manually. Proofarc is API-first:
- AI can operate it — Claude or ChatGPT can create tests, run them, analyze results
- CI/CD integration — every action is scriptable
- Reproducible — API calls are documentation
- Scalable — automate everything, no clicking