Home
arcade-dev
arcade-dev
Arcade.dev (formerly Arcade AI) is the MCP runtime for production AI agents — a platform that lets agents safely take real-world actions across SaaS apps with per-user authorization, governed tool execution, and audit logging. The Arcade Engine exposes a Tools API for listing / authorizing / executing agent tools, an Authorization API that brokers OAuth and API-key grants across 40+ providers, an OpenAI-compatible LLM endpoint, Workers and MCP Gateways for routing, Hooks for governance, and an Admin API for control-plane configuration. Arcade ships with 7,000+ pre-built integrations across Google, Microsoft, Slack, GitHub, Salesforce, HubSpot, and more, plus an MCP server framework (arcade-mcp) and SDKs in Python, TypeScript, Go, Java/Kotlin, and .NET. Available as Arcade Cloud, VPC, on-prem, or air-gapped.
9 APIs
18 Features
List, inspect, authorize, execute, and schedule Arcade tools. Tools are user-scoped agent actions addressed by fully qualified name (Toolkit.Name@version) and executed against a...
Authorize end-users against 40+ identity providers (Google, Microsoft, Slack, GitHub, Salesforce, HubSpot, and more). Issue user-scoped grants, poll authorization status, and co...
OpenAI-compatible chat completions endpoint that lets agents call LLMs through Arcade with tools attached. The engine handles tool selection, end-user authorization, and tool ex...
Register, manage, health-check, and authorize workers that host Arcade toolkits or MCP servers behind the Arcade Engine. Self-hosted workers let customers keep tool execution (a...
Create and manage MCP gateways scoped to organizations and projects. Gateways expose curated tool catalogs to MCP clients (Cursor, Claude Desktop, Claude Code, VS Code, Microsof...
Manage lifecycle hooks that fire on tool authorization, tool execution, worker health, and audit events. Hooks are the primary integration point for SIEM forwarding, audit loggi...
Administer Arcade auth providers, organization secrets, user connections, and session verification settings. The control plane behind tool authorization and end-user identity.
Manage plugins that extend the Arcade Engine with custom auth providers, transformers, and tool sources. Plugins let teams bring their own integrations behind the Arcade governa...
Operational endpoints exposing service health, engine runtime configuration, and the OpenAPI / swagger description of the Arcade Engine.
MCP runtime engine for secure, multi-user AI agent tool calling
7,000+ pre-built tool integrations across SaaS apps
Agent-optimized toolkits for Gmail, Google Calendar, Google Drive, Slack, Microsoft 365, Salesforce, HubSpot, GitHub, Jira, Asana, LinkedIn, Stripe, MongoDB, Postgres, and more
User-scoped authorization (not service accounts) — agents act as the end-user with OAuth and API-key tokens managed by Arcade
40+ auth providers including Google, Microsoft, Slack, GitHub, Salesforce, HubSpot
OpenAI-compatible chat completions endpoint with attached tools
Self-hosted workers for executing tools inside the customer perimeter
MCP gateways scoped per organization and project for MCP-client routing
Audit logs and tool-execution hooks for SIEM and governance pipelines
Tool authorization, tool execution, scheduled tools, formatted tools, and tool requirement inspection APIs
Python (arcadepy), TypeScript (arcadejs), Go, Java/Kotlin, and .NET SDKs
arcade-mcp MCP Server Framework and Tool Development library for custom toolkits
Integrations with AG2, CrewAI, Google ADK, LangChain (Python / TypeScript), Mastra, OpenAI Agents, Spring AI, TanStack AI, and Vercel AI SDK
Supported MCP clients: Cursor, Claude Desktop, Claude Code, VS Code, Microsoft Copilot Studio, GitHub Copilot
Cloud, VPC, on-premises, and air-gapped deployment options
Usage-based pricing — Hobby free, Growth $25/mo plus overage, Enterprise custom
Arcade Registry (Beta) for publishing, sharing, and monetizing agentic tools
SSO/SAML, RBAC, and tenant isolation on Enterprise
34 classes · 10 properties
JSON-LD
7 rules ·
3 errors
4 warnings
SPECTRAL
Sources
aid: arcade-dev
url: https://raw.githubusercontent.com/api-evangelist/arcade-dev/refs/heads/main/apis.yml
apis:
- aid: arcade-dev:arcade-tools-api
name: Arcade Tools API
tags:
- Tools
- Agents
- AI
- Execution
humanURL: https://docs.arcade.dev/references/api
properties:
- url: https://docs.arcade.dev/home/tools
type: Documentation
- url: openapi/arcade-tools-api-openapi.yml
type: OpenAPI
- url: json-schema/arcade-tool-schema.json
type: JSONSchema
- url: json-schema/arcade-tool-execution-schema.json
type: JSONSchema
- url: json-structure/arcade-tool-structure.json
type: JSONStructure
- url: examples/arcade-tools-list-example.json
type: Example
- url: examples/arcade-tools-execute-example.json
type: Example
- url: examples/arcade-tools-authorize-example.json
type: Example
description: >-
List, inspect, authorize, execute, and schedule Arcade tools. Tools are user-scoped agent actions addressed by
fully qualified name (Toolkit.Name@version) and executed against a stable user_id with Arcade-managed
authorization tokens.
- aid: arcade-dev:arcade-auth-api
name: Arcade Authorization API
tags:
- Authorization
- OAuth
- Identity
- Agents
humanURL: https://docs.arcade.dev/home/auth
properties:
- url: https://docs.arcade.dev/home/auth
type: Documentation
- url: openapi/arcade-auth-api-openapi.yml
type: OpenAPI
- url: json-schema/arcade-authorization-schema.json
type: JSONSchema
- url: json-structure/arcade-authorization-structure.json
type: JSONStructure
- url: examples/arcade-auth-authorize-example.json
type: Example
- url: examples/arcade-auth-status-example.json
type: Example
description: >-
Authorize end-users against 40+ identity providers (Google, Microsoft, Slack, GitHub, Salesforce, HubSpot, and
more). Issue user-scoped grants, poll authorization status, and confirm users so agents can call tools as the
end-user rather than a shared service account.
- aid: arcade-dev:arcade-llm-api
name: Arcade LLM API
tags:
- LLM
- AI
- Chat
- OpenAI Compatible
humanURL: https://docs.arcade.dev/home/chat-overview
properties:
- url: https://docs.arcade.dev/home/chat-overview
type: Documentation
- url: openapi/arcade-llm-api-openapi.yml
type: OpenAPI
- url: examples/arcade-llm-chat-completions-example.json
type: Example
description: >-
OpenAI-compatible chat completions endpoint that lets agents call LLMs through Arcade with tools attached. The
engine handles tool selection, end-user authorization, and tool execution so agents need only speak the OpenAI
Chat Completions protocol.
- aid: arcade-dev:arcade-workers-api
name: Arcade Workers API
tags:
- Workers
- MCP
- Infrastructure
- Agents
humanURL: https://docs.arcade.dev/home/workers
properties:
- url: https://docs.arcade.dev/home/workers
type: Documentation
- url: openapi/arcade-workers-api-openapi.yml
type: OpenAPI
- url: json-schema/arcade-worker-schema.json
type: JSONSchema
- url: json-structure/arcade-worker-structure.json
type: JSONStructure
- url: examples/arcade-workers-create-example.json
type: Example
description: >-
Register, manage, health-check, and authorize workers that host Arcade toolkits or MCP servers behind the Arcade
Engine. Self-hosted workers let customers keep tool execution (and end-user tokens) inside their own perimeter.
- aid: arcade-dev:arcade-gateways-api
name: Arcade Gateways API
tags:
- Gateways
- MCP
- Agents
humanURL: https://docs.arcade.dev/guides/mcp-gateways
properties:
- url: https://docs.arcade.dev/guides/mcp-gateways
type: Documentation
- url: openapi/arcade-gateways-api-openapi.yml
type: OpenAPI
- url: examples/arcade-gateways-create-example.json
type: Example
description: >-
Create and manage MCP gateways scoped to organizations and projects. Gateways expose curated tool catalogs to MCP
clients (Cursor, Claude Desktop, Claude Code, VS Code, Microsoft Copilot Studio, GitHub Copilot).
- aid: arcade-dev:arcade-hooks-api
name: Arcade Hooks API
tags:
- Hooks
- Webhooks
- Events
- Agents
humanURL: https://docs.arcade.dev/guides/audit-logs
properties:
- url: https://docs.arcade.dev/guides/audit-logs
type: Documentation
- url: openapi/arcade-hooks-api-openapi.yml
type: OpenAPI
- url: json-schema/arcade-hook-schema.json
type: JSONSchema
- url: examples/arcade-hooks-create-example.json
type: Example
description: >-
Manage lifecycle hooks that fire on tool authorization, tool execution, worker health, and audit events. Hooks are
the primary integration point for SIEM forwarding, audit logging, and downstream governance pipelines.
- aid: arcade-dev:arcade-admin-api
name: Arcade Admin API
tags:
- Administrative
- Secrets
- Identity
- Agents
humanURL: https://docs.arcade.dev/references/api
properties:
- url: https://docs.arcade.dev/home/auth-providers
type: Documentation
- url: openapi/arcade-admin-api-openapi.yml
type: OpenAPI
- url: examples/arcade-admin-secrets-set-example.json
type: Example
description: >-
Administer Arcade auth providers, organization secrets, user connections, and session verification settings. The
control plane behind tool authorization and end-user identity.
- aid: arcade-dev:arcade-plugins-api
name: Arcade Plugins API
tags:
- Plugins
- Agents
- Extensibility
humanURL: https://docs.arcade.dev/references/api
properties:
- url: openapi/arcade-plugins-api-openapi.yml
type: OpenAPI
description: >-
Manage plugins that extend the Arcade Engine with custom auth providers, transformers, and tool sources. Plugins
let teams bring their own integrations behind the Arcade governance and auth layer.
- aid: arcade-dev:arcade-operations-api
name: Arcade Operations API
tags:
- Operations
- Health
- Configuration
humanURL: https://docs.arcade.dev/references/api
properties:
- url: openapi/arcade-operations-api-openapi.yml
type: OpenAPI
- url: examples/arcade-operations-health-example.json
type: Example
description: >-
Operational endpoints exposing service health, engine runtime configuration, and the OpenAPI / swagger description
of the Arcade Engine.
common:
- type: Portal
url: https://arcade.dev
- type: Documentation
url: https://docs.arcade.dev
- type: GettingStarted
url: https://docs.arcade.dev/home
- type: SignUp
url: https://api.arcade.dev/dashboard/auth/login
- type: Pricing
url: https://arcade.dev/pricing
- type: Blog
url: https://arcade.dev/blog
- type: ChangeLog
url: https://docs.arcade.dev/home/release-notes
- url: https://github.com/ArcadeAI
type: GitHubOrganization
- url: https://github.com/ArcadeAI/arcade-py
name: Python SDK (arcadepy)
type: SDK
- url: https://github.com/ArcadeAI/arcade-js
name: TypeScript / JavaScript SDK
type: SDK
- url: https://github.com/ArcadeAI/arcade-go
name: Go SDK
type: SDK
- url: https://github.com/ArcadeAI/arcade-java
name: Java / Kotlin SDK
type: SDK
- url: https://github.com/ArcadeAI/arcade-dotnet
name: .NET SDK
type: SDK
- url: https://github.com/ArcadeAI/arcade-mcp
name: arcade-mcp — MCP Server Framework and Tool Development library
type: SDK
- url: https://github.com/ArcadeAI/agent-templates
name: Agent Templates
type: CodeExamples
- url: https://github.com/ArcadeAI/blueprint-mcp
name: blueprint-mcp (codebase diagramming MCP server)
type: Tool
- url: https://github.com/ArcadeAI/agent-library
name: agent-library (personal knowledge MCP)
type: Tool
- url: https://github.com/ArcadeAI/safeword
name: safeword — personal AI agent guides and templates
type: CodeExamples
- url: https://github.com/ArcadeAI/docs
name: Arcade Documentation Source
type: Documentation
- url: https://modelcontextprotocol.io
name: Model Context Protocol
type: Documentation
- url: https://docs.arcade.dev/references/api
type: Documentation
name: API Reference
- url: https://docs.arcade.dev/home/install/cli
type: SDK
name: arcade CLI
- url: https://docs.arcade.dev/home/auth-providers
type: Documentation
name: Auth Providers (40+)
- url: https://docs.arcade.dev/home/tools
type: Documentation
name: Pre-built Tools (7,000+ integrations)
- url: https://docs.arcade.dev/home/build-tools
type: Documentation
name: Build Custom Tools
- url: https://docs.arcade.dev/guides/mcp-gateways
type: Documentation
name: MCP Gateways
- url: https://docs.arcade.dev/guides/audit-logs
type: Documentation
name: Audit Logs
- url: https://docs.arcade.dev/guides/contextual-access
type: Documentation
name: Contextual Access
- url: https://docs.arcade.dev/guides/deployment
type: Documentation
name: Deployment (Cloud / VPC / On-Prem / Air-Gapped)
- url: https://docs.arcade.dev/home/security
type: Documentation
name: Security Program
- url: https://arcade.dev/registry
type: Documentation
name: Arcade Registry (Beta)
- url: https://discord.gg/arcadedev
type: Forum
- url: https://x.com/TryArcade
type: X
- url: plans/arcade-dev-plans-pricing.yml
type: Plans
- url: rate-limits/arcade-dev-rate-limits.yml
type: RateLimits
- url: finops/arcade-dev-finops.yml
type: FinOps
- url: rules/arcade-dev-rules.yml
type: SpectralRules
- url: json-ld/arcade-dev-context.jsonld
type: JSONLD
- url: vocabulary/arcade-dev-vocabulary.yml
type: Vocabulary
- type: Features
data:
- MCP runtime engine for secure, multi-user AI agent tool calling
- 7,000+ pre-built tool integrations across SaaS apps
- >-
Agent-optimized toolkits for Gmail, Google Calendar, Google Drive, Slack, Microsoft 365, Salesforce, HubSpot,
GitHub, Jira, Asana, LinkedIn, Stripe, MongoDB, Postgres, and more
- >-
User-scoped authorization (not service accounts) — agents act as the end-user with OAuth and API-key tokens
managed by Arcade
- 40+ auth providers including Google, Microsoft, Slack, GitHub, Salesforce, HubSpot
- OpenAI-compatible chat completions endpoint with attached tools
- Self-hosted workers for executing tools inside the customer perimeter
- MCP gateways scoped per organization and project for MCP-client routing
- Audit logs and tool-execution hooks for SIEM and governance pipelines
- Tool authorization, tool execution, scheduled tools, formatted tools, and tool requirement inspection APIs
- Python (arcadepy), TypeScript (arcadejs), Go, Java/Kotlin, and .NET SDKs
- arcade-mcp MCP Server Framework and Tool Development library for custom toolkits
- >-
Integrations with AG2, CrewAI, Google ADK, LangChain (Python / TypeScript), Mastra, OpenAI Agents, Spring AI,
TanStack AI, and Vercel AI SDK
- 'Supported MCP clients: Cursor, Claude Desktop, Claude Code, VS Code, Microsoft Copilot Studio, GitHub Copilot'
- Cloud, VPC, on-premises, and air-gapped deployment options
- Usage-based pricing — Hobby free, Growth $25/mo plus overage, Enterprise custom
- Arcade Registry (Beta) for publishing, sharing, and monetizing agentic tools
- SSO/SAML, RBAC, and tenant isolation on Enterprise
sources:
- https://arcade.dev
- https://arcade.dev/pricing
- https://docs.arcade.dev
- https://api.arcade.dev/v1/swagger
- https://github.com/ArcadeAI
updated: '2026-05-25'
created: '2026-05-25'
modified: '2026-05-25'
position: Producing
description: >-
Arcade.dev (formerly Arcade AI) is the MCP runtime for production AI agents — a platform that lets agents safely take
real-world actions across SaaS apps with per-user authorization, governed tool execution, and audit logging. The
Arcade Engine exposes a Tools API for listing / authorizing / executing agent tools, an Authorization API that brokers
OAuth and API-key grants across 40+ providers, an OpenAI-compatible LLM endpoint, Workers and MCP Gateways for
routing, Hooks for governance, and an Admin API for control-plane configuration. Arcade ships with 7,000+ pre-built
integrations across Google, Microsoft, Slack, GitHub, Salesforce, HubSpot, and more, plus an MCP server framework
(arcade-mcp) and SDKs in Python, TypeScript, Go, Java/Kotlin, and .NET. Available as Arcade Cloud, VPC, on-prem, or
air-gapped.
maintainers:
- FN: Kin Lane
email: info@apievangelist.com
X: apievangelist
url: https://apievangelist.com
specificationVersion: '0.16'