AGENTKIT V2 IS LIVE

The framework for resilient agents.

The framework for resilient agents.

The framework for resilient agents.

Agentkit is design-first infrastructure for AI agents. Build complex, long-running agentic workflows with observability, memory, and multi-model support built-in.

AGENT-LIVE
● Running
agent.run() · lifecyclev2.0

Built for long-running execution.

Built for long-running execution.

Connect to any API, database, or service. Define custom tools with full type safety and auto-generated schemas.

Connect to any API, database, or service. Define custom tools with full type safety and auto-generated schemas.

AGENT-RT-001
Running
runtime.ts

Auto-healing Loops

If an agent hits an error or hallucinates a tool call, the runtime automatically rolls back and prompts for correction.

Async State

Save and resume agent state across different sessions or processes.

Edge Compatible

Run your agents on Vercel Edge, AWS Lambda, or local servers with no overhead.

Auto-healing Loops

If an agent hits an error or hallucinates a tool call, the runtime automatically rolls back and prompts for correction.

Async State

Save and resume agent state across different sessions or processes.

Edge Compatible

Run your agents on Vercel Edge, AWS Lambda, or local servers with no overhead.

40+, integrations custom

40+, integrations custom

A comprehensive suite designed specifically for the complex workflows of modern legal and compliance professionals.

A comprehensive suite designed specifically for the complex workflows of modern legal and compliance professionals.

AGENT-TL-002
Ready
tools.ts

Pre-built Integrations

Slack, Discord, Notion, GitHub, Stripe, PostgreSQL, Redis, and 30+ more ready to use.

Type-Safe Custom Tools

Define your own tools with Zod schemas. Full IntelliSense and compile-time validation.

MCP Server Support

Native Model Context Protocol support. Connect any MCP server and your agents get new tools at runtime — no custom adapters, no redeploys.

Pre-built Integrations

Slack, Discord, Notion, GitHub, Stripe, PostgreSQL, Redis, and 30+ more ready to use.

Type-Safe Custom Tools

Define your own tools with Zod schemas. Full IntelliSense and compile-time validation.

MCP Server Support

Native Model Context Protocol support. Connect any MCP server and your agents get new tools at runtime — no custom adapters, no redeploys.

Persistent context no extra DB.

Persistent context no extra DB.

Built-in memory management with automatic context windowing, summaries, and vector search. No external database required.

Built-in memory management with automatic context windowing, summaries, and vector search. No external database required.

AGENT-MM-003
Active
memory · session #4827

Automatic Summaries

Long conversations are compressed into running summaries to preserve context without hitting token limits.

Vector Recall

Semantic search over past interactions. Retrieve relevant memories even from distant conversation history.

Zero Config

Memory persists automatically. Configure TTL, encryption, and storage backends with one line.

Automatic Summaries

Long conversations are compressed into running summaries to preserve context without hitting token limits.

Vector Recall

Semantic search over past interactions. Retrieve relevant memories even from distant conversation history.

Zero Config

Memory persists automatically. Configure TTL, encryption, and storage backends with one line.

Full observability for every run.

Full observability for every run.

See every thought, tool call, and decision your agent makes. Debug production issues with complete visibility.

See every thought, tool call, and decision your agent makes. Debug production issues with complete visibility.

AGENT-TR-004
Live
trace · run #4827

Step-by-Step Execution Log

Every LLM call, tool execution, and reasoning step is logged with timing, tokens, and cost breakdowns.

Real-time Streaming

Watch agents work live. Stream traces to your dashboard as they happen.

Cost Tracking

Per-run cost attribution. Track spending by agent, user, or organization with detailed breakdowns.

Step-by-Step Execution Log

Every LLM call, tool execution, and reasoning step is logged with timing, tokens, and cost breakdowns.

Real-time Streaming

Watch agents work live. Stream traces to your dashboard as they happen.

Cost Tracking

Per-run cost attribution. Track spending by agent, user, or organization with detailed breakdowns.

Production-safe by default.

Production-safe by default.

Block prompt injection, mask sensitive data, and cap runaway budgets. Security primitives baked into the runtime, not bolted on after launch.

Block prompt injection, mask sensitive data, and cap runaway budgets. Security primitives baked into the runtime, not bolted on after launch.

AGENT-SG-001
Filtering
guard.ts · prompt firewall

Prompt Shield

Detect and block prompt injection attacks before they reach the model.

PII Masking

Automatically strip sensitive data from logs and third-party API calls.

Cost Guardrails

Set budget limits per agent run or per session to prevent infinite loops.

Prompt Shield

Detect and block prompt injection attacks before they reach the model.

PII Masking

Automatically strip sensitive data from logs and third-party API calls.

Cost Guardrails

Set budget limits per agent run or per session to prevent infinite loops.

Questions? Answers.

Questions? Answers.

How runs work, what you'll pay, and what happens when you scale. / No surprises. No footnotes. Just the answers that matter.

How runs work, what you'll pay, and what happens when you scale. / No surprises. No footnotes. Just the answers that matter.

A run is one complete agent execution — from the initial prompt to the final response, including any tool calls, retries, and model invocations along the way.
Nothing breaks. You'll be charged a per-run overage rate ($0.002/run on Pro). We'll notify you at 80% and 100% of your limit so there are no surprises.
No. If a run fails due to an Agentkit error, it's free. You only pay for successful agent executions that return a result.
Yes — on Pro and Custom plans you can connect your own OpenAI, Anthropic, Gemini, or self-hosted endpoints. You'll only be charged for Agentkit infrastructure, not model tokens.
Yes. 1,000 runs per month, every month, no credit card required. It's a permanent free tier, not a trial.
Early-stage startups (under $5M raised) get 50% off Pro for 12 months. Open-source maintainers get Pro free. Email startups@agentkit.dev to apply.
On-prem and private VPC deployment are available on the Custom plan. The core runtime is also open-source and can run on your own infrastructure.
SSO/SAML, dedicated infrastructure, custom SLAs, audit logs, a named support engineer, on-prem deployment, and volume pricing on runs.
Yes. Cancel from your dashboard in one click. Monthly plans stop at the end of the billing period. Annual plans are refunded pro-rata for unused months.

Start free, pay for what runs. No seats, no platform fees, no surprise overages — just runs, tokens, and the tools you actually use.

Start free, pay for what runs. No seats, no platform fees, no surprise overages — just runs, tokens, and the tools you actually use.

The framework for building, deploying, and observing production AI agents. Made in Berlin, shipped globally.

© AgentKit Inc.
Berlin — 2026

The framework for building, deploying, and observing production AI agents. Made in Berlin, shipped globally.

© AgentKit Inc.
Berlin — 2026

The framework for building, deploying, and observing production AI agents. Made in Berlin, shipped globally.

© AgentKit Inc.
Berlin — 2026

Create a free website with Framer, the website builder loved by startups, designers and agencies.