Website · Docs · Changelog · Blog · Slack · X · LinkedIn · Careers
Speakeasy is the AI control plane for the AI-native enterprise. It sits between every AI agent (Claude, ChatGPT, Cursor, Copilot, Codex, internal agents) and the systems those agents need to reach: SaaS tools, internal APIs, and databases. Every prompt, response, and tool call flows through one governed path.
| Connect | One catalog of MCP servers, SaaS connectors, and internal APIs. Per-team registries, identity from your IdP, one URL for every agent. |
| Secure | Inspect every prompt, response, and tool call. Block PII and data exfiltration. Detect prompt injection and shadow MCP servers. |
| Control | Scoped access by team, role, or tool. OAuth 2.1 with PKCE and DCR, even when upstream servers lack it. Policies with a full audit trail. |
| Observe | Token use and adoption by team, client, tool, and user. Integrates with the SIEM and incident response tooling already in place. |
SOC 2 Type II. Integrates with Okta, Microsoft Entra ID, Auth0, WorkOS, Google Workspace, and Ping Identity.
| Repository | What it is |
|---|---|
gram |
The open source core of the Speakeasy AI control plane: MCP hosting, gateway, and toolsets. |
speakeasy |
The Speakeasy CLI. Generate SDKs, MCP servers, Terraform providers, and CLIs from OpenAPI. |
skills |
Official agent skills for working with Speakeasy from Claude Code, Codex, Cursor, and others. |
docs-mcp |
Index Markdown docs into an MCP server that agents can search and cite. |
# Give your coding agent the Speakeasy skills
npx skills add speakeasy-api/skills
# Install the CLI and generate from an OpenAPI spec
brew install speakeasy-api/tap/speakeasy
speakeasy quickstartPoint an agent at the Platform MCP server, https://app.getgram.ai/platform-mcp, to register MCP servers, distribute them to teams, and query usage from chat. Setup for each agent is in agents.md.
