ā†back to changelog

Introducing ChatBotKit Agent SDK

We're thrilled to announce the public launch of the ChatBotKit Agent SDK - a powerful new toolkit for building autonomous AI agents that can use custom tools and execute complex, multi-step tasks with the full power of the ChatBotKit platform.

What is the Agent SDK?

The Agent SDK enables you to create AI agents that go beyond simple conversations. These agents can interact with your local environment, call external APIs, execute workflows, and leverage ChatBotKit's extensive platform integrations and features - all while maintaining autonomous control over task execution.

Key Features

Custom Tool Integration

Define type-safe tools using Zod schemas that your agent can use to interact with any system:

Tools are automatically exposed to the AI with proper validation, error handling, and type checking - so you can focus on building functionality rather than wiring infrastructure.

Dual Execution Modes

Complete Mode - Stream agent responses with tool execution for conversational AI that can take actions:

Execute Mode - Run agents with built-in planning, progress tracking, and controlled exit for autonomous task completion:

Built-in System Tools

Execute mode provides three essential system tools that enable sophisticated task management:

  • plan - Create or update task execution plans with clear, actionable steps
  • progress - Track completion status, identify blockers, and report on current work
  • exit - Signal task completion with status codes (0 for success, non-zero for failure)

These tools give your agent the ability to think strategically, break down complex problems, and communicate its progress - just like a human developer would.

Full Platform Integration

Agents run with complete ChatBotKit platform capabilities:

  • Authenticated integrations - Access Slack, Discord, Jira, Linear, Google Workspace, Notion, and more without managing OAuth flows
  • Dataset connections - Query your knowledge bases and retrieve relevant information
  • Skillsets and abilities - Leverage pre-built capabilities from the platform
  • Custom functions - Combine platform functions with your local tools

This is where the real power emerges: your agent can bridge local operations (file system access, git commands, system metrics) with remote collaboration tools (team notifications, ticket creation, data logging).

Real-World Use Case: Development Monitor

Here's an example that showcases the Agent SDK's unique capabilities - monitoring local development and syncing insights to remote platforms:

In this example, the agent:

  1. Monitors local state - Git changes, dependencies, file system (can't be accessed remotely)
  2. Analyzes and decides - Evaluates if changes are production-ready
  3. Creates Jira tickets - Via ChatBotKit's Jira integration (no API setup needed)
  4. Sends Slack alerts - Team notifications through authenticated ChatBotKit channels

ChatBotKit handles all remote authentication, rate limiting, and API complexity - you just define the local tools and let the agent orchestrate everything.

Advanced Capabilities

Parallel Tool Execution

The SDK automatically handles parallel tool execution when appropriate, maximizing efficiency without requiring manual orchestration.

Type-Safe Tool Definitions

Using Zod schemas ensures your tools are properly validated before execution, catching errors early and providing clear feedback to the AI agent.

Streaming Events

Both execution modes provide rich event streams so you can track:

  • Tool call start/end events
  • Tool execution errors
  • Iteration progress
  • Task completion status
  • Token streaming for responses

Error Handling

Built-in error handling ensures that tool failures are gracefully communicated to the agent, allowing it to adapt its strategy and continue working toward the goal.

Getting Started

Install the Agent SDK alongside the ChatBotKit SDK:

You'll need a ChatBotKit API token to get started. Create one in your ChatBotKit dashboard.

Documentation

For comprehensive documentation including API references, advanced examples, and best practices, visit:

Use Cases

The Agent SDK opens up countless possibilities for business automation and workflow optimization:

Development & DevOps

  • CI/CD automation - Monitor build pipelines, analyze test results, create tickets for failures, notify teams
  • Code review workflows - Analyze pull requests, run security scans, update Jira, post to Slack
  • System monitoring - Track server metrics, detect anomalies, create incident tickets, alert on-call teams

Business Operations

  • Customer onboarding - Process applications, verify documents, update CRM, schedule follow-ups
  • Invoice processing - Extract data from PDFs, validate against orders, update accounting systems, send notifications
  • Expense management - Scan receipts, categorize expenses, update spreadsheets, flag anomalies for review
  • Contract management - Monitor expiration dates, extract key terms, update tracking systems, notify stakeholders

Sales & Marketing

  • Lead qualification - Analyze incoming leads, enrich with external data, score prospects, route to sales teams
  • Campaign monitoring - Track performance metrics, generate reports, update dashboards, alert on threshold breaches
  • Content distribution - Generate content variations, publish to multiple platforms, track engagement, compile analytics

Customer Support

  • Ticket triage - Analyze support requests, categorize issues, assign priority, route to specialists
  • Knowledge base management - Monitor common questions, generate FAQ entries, update documentation, sync to CMS
  • Escalation workflows - Detect critical issues, gather context, create high-priority tickets, notify managers

Data & Analytics

  • Report automation - Collect data from multiple sources, generate visualizations, compile reports, distribute via email/Slack
  • Data synchronization - Extract from local databases, transform data, sync to cloud platforms, verify integrity
  • Analytics pipelines - Aggregate metrics, perform calculations, update dashboards, alert on trends

Anywhere you need to bridge local operations with remote collaboration tools, the Agent SDK provides the perfect solution for automating repetitive business processes while maintaining human oversight where needed.

What Makes This Different?

Unlike traditional automation tools or simple chatbots, the Agent SDK combines:

  1. Autonomous decision-making - Agents plan, execute, and adapt without constant supervision
  2. Local + remote capabilities - Access local systems AND authenticated remote services
  3. Type-safe tools - Zod validation ensures reliability
  4. Platform integration - Leverage 30+ integrations without API configuration
  5. Flexible execution - Choose between conversational or task-oriented modes

It's the best of both worlds: the autonomy of AI agents with the power of the ChatBotKit platform.

What's Next

We're excited to see what you build with the Agent SDK! This is just the beginning - we're already working on additional features including:

  • More system tools for common workflows
  • Enhanced debugging and monitoring capabilities
  • Agent collaboration patterns
  • Pre-built agent templates for common use cases

We'd love to hear your feedback and see what you create. Share your agent implementations, ask questions, and report any issues through our or on GitHub.

SDKagentsAIdevelopers