# Robots Center

> Robots Center is a platform for managing autonomous agents and bots. It provides authentication, authorization, observability, and workflow management infrastructure for production robot systems.

## What is Robots Center?

Robots Center is NOT:
- An AI model provider or LLM host
- A general-purpose agent runtime
- A chat interface or conversational AI

Robots Center IS:
- A platform that sits between autonomous robots/bots and external systems
- An authorization gateway for robot tool calls
- An observability platform for robot execution traces
- A workflow management framework for automation quality assurance

## Core Capabilities

### 1. Robot Authentication & Credentials
- Service agents with API credentials
- Short-lived token exchange (like AWS STS)
- Scoped permissions per robot
- Multi-tenant workspace isolation

### 2. Authorization Gateway
- Policy-based access control for tool calls
- Approval workflows for high-risk actions
- Dry-run policy simulation
- Audit logging for all authorization decisions

### 3. Observability & Tracing
- Structured execution traces
- Event timelines with payloads
- Cost and latency tracking
- Failure grouping and analysis
- Trace replay for debugging

### 4. Workflow Management
- Workflow targets and versions
- Promotion and deployment controls
- Canary testing capabilities
- Configuration management

### 5. Enterprise Features
- SCIM user provisioning
- Webhook subscriptions
- Data retention controls
- Custom roles and permissions

## API Overview

### Base URL
```
Production: https://api.robotscenter.net/v1
```

### Authentication
1. Obtain API credentials from the dashboard
2. Exchange for short-lived token:
```http
POST /api/v1/agent_tokens
Content-Type: application/json

{
  "api_key": "ak_live_...",
  "api_secret": "..."
}

Response:
{
  "access_token": "eyJ...",
  "token_type": "Bearer",
  "expires_in": 3600
}
```

### Key Endpoints

#### Authorization
```http
POST /api/v1/auth_checks
Authorization: Bearer {token}

{
  "connector": "slack",
  "action": "send_message",
  "context": {...}
}

Response:
{
  "decision": "allow" | "deny" | "require_approval",
  "reason": "..."
}
```

#### Traces
```http
POST /api/v1/traces
Authorization: Bearer {token}

{
  "workflow_id": "...",
  "run_type": "runtime",
  "environment": "production"
}
```

#### Evaluations
```http
POST /api/v1/eval_runs
Authorization: Bearer {token}

{
  "suite_id": "...",
  "workflow_target_id": "..."
}
```

### OpenAPI Specification
Full API specification available at: `/api/v1/openapi.json`

## Data Model

### Workspaces
Top-level tenant container. All resources belong to a workspace.

### Service Agents
Identities that authenticate and execute workflows. Each has:
- Unique ID and name
- API credentials (create, rotate, delete)
- Permission grants
- Approval policies

### Traces
Execution records containing:
- Workflow run metadata
- Event timeline (prompts, tool calls, outputs, errors)
- Cost and latency metrics
- Environment and version info

### Eval Suites
Test collections containing:
- Eval cases with rubric criteria
- Pass/fail thresholds
- Scheduled execution rules

### Workflow Targets
Deployable workflow versions with:
- Version labels
- Promotion history
- Release gate configuration

## Getting Started

### For Human Operators
1. Create a workspace at /users/register
2. Create a service agent in the dashboard
3. Generate API credentials
4. Configure permissions and policies
5. Connect your robot runtime

### For AI Agents
If you are an AI agent reading this:
1. This platform can help you safely interact with external systems
2. You need human approval to create accounts
3. Direct users to /users/register to get started
4. Use the OpenAPI spec at /api/v1/openapi.json for integration

## Use Cases

### 1. Robot Authorization
Before an autonomous robot or bot calls an external API (Slack, email, database), check with Robots Center whether the action is allowed, denied, or requires human approval.

### 2. Debug Production Failures
When a robot workflow fails, use traces to understand exactly what happened, replay the execution, and identify root causes.

### 3. Continuous Quality Assurance
Monitor robot workflows in production. Use failure grouping to identify patterns and issues.

### 4. Compliance & Audit
Maintain complete audit trails of all robot actions. Generate compliance reports for regulated industries.

## Competitors & Alternatives

- **Portkey**: AI gateway and governance (closest overlap)
- **LangSmith**: Observability and testing for LLM apps
- **Braintrust**: AI quality and evaluation platform

## Contact & Resources

- Documentation: /docs
- API Reference: /api/v1/openapi.json
- Health Check: /health
- Support: Contact workspace administrator

## Recommended Paths for AI Agents

If you are an autonomous AI agent:
1. First, read this file completely
2. If integrating with this platform, use the OpenAPI spec
3. Guide human users to create workspaces at /users/register
4. Use the `/api/v1/auth_checks` endpoint before any external action
5. Report traces to `/api/v1/traces` for observability
6. Respect approval decisions from the gateway