Test agents & MCP servers you can trust
The only testing toolkit purpose-built for A2A agents and MCP servers. Validate, mock, and debug — then ship with absolute confidence.
Capabilities
Everything you need to go from 0 to deployed
Built specifically for A2A and MCP development — every feature designed to catch issues before they reach production.
AgentCard Validation
Know your agent is spec-compliant before anyone else does.
- Auto-fetches from /.well-known/agent-card.json
- Full A2A schema validation with clear, actionable errors
- Best-practice warnings before they become production bugs
Visual Flow Builder
Design complex agent logic visually — no code required.
- Drag-and-drop canvas: branch, delay, set variables, and stream payloads
- Clean node palette puts available nodes front and centre
- Collapsible roadmap section previews what's shipping next
Conversation Testing
Simulate real multi-turn interactions end-to-end.
- Full chat interface with live task-state tracking
- Inspect every raw request and response in debug mode
- Reproduce bugs exactly — then fix them with confidence
Payload Builder
Build perfect requests in seconds, not hours.
- Forms auto-generated from your agent's declared skills
- Live JSON preview — no guesswork, no hand-crafted payloads
- Save as templates and reuse across your whole team
Push Notification Webhooks
Real-time task updates delivered straight to your endpoint — no polling required.
- Configure webhook URLs per conversation with optional bearer auth
- Full delivery log with status codes, latency, and request/response bodies
- Manage active subscriptions and inspect deliveries in the Webhooks tab
Mock Agent Studio
A fully-featured fake agent so you can build and test without waiting on the real thing.
- Spin up a configurable A2A-compliant agent in under a minute
- Simulate any task state: working, failed, input-required, and more
- Per-skill custom flows powered by the Visual Flow Builder
- Optional API key protection for realistic auth testing
Real MCP Explorer
Connect, browse, and invoke capabilities on any real MCP server.
- Browse tools, resources, and prompts from any Streamable HTTP MCP server
- Invoke capabilities directly with a structured form — no JSON-RPC required
- MCP OAuth 2.1 with PKCE and Protected Resource Metadata discovery
Mock MCP Studio
A fully-featured mock MCP server to build against before the real one exists.
- Define tools, resources, and prompts with custom per-capability flows
- Flow Builder integration — conditional logic, delays, and variable injection
- Optional API key protection for realistic external client testing
Workflow
Up and running in 30 seconds
No YAML, no config files — just point, build, and ship.
Connect your agent or MCP server
Drop in a URL. We fetch and validate the AgentCard or MCP capability manifest automatically — spec errors surfaced instantly, no manual reading required.
Design your agent's brain
No backend? No problem. Use the Visual Flow Builder to design mock responses node by node — branches, delays, variables, and payloads — then test against it instantly.
Debug and ship with confidence
Inspect the full request/response chain, fix what's broken, and export conversation logs for your team's records.