Skill Server Integration
The Skill Server integration is a simpler, agent-first way to expose your ChatBotKit skillsets. It is the sibling of the MCP Server integration: the same skillset, but instead of speaking the Model Context Protocol, a Skill Server is a plain HTTP endpoint that any agent, script, or curl command can drive.
A Skill Server exposes a single authenticated URL, dispatched by HTTP method. There is no client library, no JSON-RPC envelope, and no handshake to manage.
What You Can Do
- Read a manual: A
GETreturns a concise, text-first description of every ability in the linked skillset and how to call it, generated from the skillset so it always reflects what is callable. - Invoke abilities: A
POSTcalls an ability by name with its input. - Use any client: Works with any agent framework, automation tool, shell script, or cron job that can make an HTTP request.
- Choose your format: Responses are plain text by default, with structured JSON available via a
?format=jsonquery parameter or anAccept: application/jsonheader.
How It Works
The whole protocol is two HTTP methods on one URL. An agent reads the manual once with a GET, discovers the available abilities and their inputs, then calls them directly with POST requests. Because the manual is generated from the skillset itself, it stays accurate as you change abilities. Text-first responses are token-cheap and resilient to context truncation, which suits agents reading and acting in a loop.
Setup
Create a Skill Server integration in ChatBotKit and link it to the skillset you want to expose. The integration provides an authenticated URL. Point your agent or script at it, issue a GET to read the manual, and POST to invoke abilities.
Practical Uses
Skill Servers are ideal for connecting ChatBotKit abilities to external agent frameworks, automation pipelines, and scripts without writing integration code. If a tool can make an HTTP request, it can use a Skill Server to search datasets, call APIs, and trigger the actions your skillset defines.