Kastra

For AI teams: Kastra enforces authorization policies on Claude, Cursor, and OpenAI agents in sub-1ms so unauthorized tool use and prompt injection never reach production.

Artificial Intelligence · Developer Tools · Freemium

Visit website

In category: #116 of 2034 · Artificial Intelligence

Product Hunt listing

Listing status

398upvotes

Community score

—

Founder

Compared against 6 alternatives

Market comparison

Launched on Launch Llama Jul 28, 2026 · #21 of 67 that day

About

Kastra is the runtime authorization layer for AI agents. It decides what agents can and cannot do before actions execute, enforcing policies with sub-1 ms latency across tools, prompts, inputs, and outputs. Use one control plane to govern agents and policies across Claude Code, Cursor, Codex, OpenClaw, the Anthropic SDK, the OpenAI SDK, and more. Prevent unauthorized tool use, prompt injection, and exposure of sensitive data before they become incidents. Trust the rules, not the agents.

Launch Llama's Take

Based on Launch Llama data as of November 2026

Kastra has 398 upvotes, including the votes it received on Product Hunt, which ranks it #116 of 2,034 Artificial Intelligence tools on Launch Llama by upvotes, in the top 10%. It has been listed on Launch Llama since July 2026. No Launch Llama members have upvoted it yet and it has no reviews yet.

Kastra is the runtime authorization layer for AI agents, deciding what they can do before actions execute. Unlike AI-FW, which sits between AI agents and LLM backends to govern traffic, Kastra authorizes prompts, tool calls, shell commands, and API requests.

Kastra vs AI-FW

Kastra vs AI-FW: side-by-side comparison
	Kastra	AI-FW	AgentsKit	ranwhat
What it does	For AI teams: Kastra enforces authorization policies on Claude, Cursor, and OpenAI agents in sub-1ms so unauthorized tool use and prompt…	Helps platform and security teams govern AI agent traffic to	For developers: AgentsKit provides open-source TypeScript components for building AI agents so you avoid vendor lock-in and swap layers f…	Open-source AI agent security: an audit trail of what Claude
Pricing	Freemium	Freemium	Free	Not stated
Free plan	Yes, free plan	Yes, free plan	Yes, free to use	Not stated
Category	Developer Tools	Artificial Intelligence	Artificial Intelligence	AI Agent
Upvotes on Launch Llama	398	2	43	2
Listed on Launch Llama	July 2026	October 2026	August 2026	October 2026

See all Kastra alternatives

Key Features

Evaluate policies and return allow, deny, redact, or hold decisions
Authorize prompts, tool calls, shell commands, and API requests
Create cryptographically chained audit trails of all decisions
Govern Claude Code, Cursor, Codex, and other AI tools
Connect via HTTP API, Edge installation, or MCP client

Use Cases

Teams deploying agents that take real-world actions and need to control what those agents can do
Organizations preventing unauthorized tool use and prompt injection attacks

Who it is for

AI teams running production agents that take real-world actions such as tool calls, database operations, and API requests.
Developers using Claude Code, Cursor, or Codex who need policy enforcement on their local machines.
Operators and auditors who need to explain and verify AI agent permission decisions.

Pricing

Pro, Team, and Enterprise plans available. 15-day Pro trial requires no credit card.

View Kastra pricing →

Kastra at a glance

01What it is: Kastra: For AI teams: Kastra enforces authorization policies on Claude, Cursor, and OpenAI agents in sub-1ms so unauthorized tool use and prompt injection never reach production.
02Pricing: Kastra pricing: Pro, Team, and Enterprise plans available.
03Best for: Kastra is best for aI teams running production agents that take real-world actions such as tool calls, database operations, and API requests., developers using Claude Code, Cursor, or Codex who need policy enforcement on their local machines. and operators and auditors who need to explain and verify AI agent permission decisions.
04Alternatives: Top Kastra alternatives on Launch Llama: AgentsKit, ranwhat and Kavela.

Compare Kastra with AgentsKit

FAQ

What is AI execution governance?

AI execution governance is a security approach that requires explicit authorization before any AI workload can execute. It ensures every AI action is approved by policy, logged for audit, and controlled in real-time.

How does Kastra work?

Kastra sits between your AI workloads and their execution targets. Every request is evaluated against your policies in sub-millisecond time, and only authorized actions proceed. All decisions are logged with full context.

What clouds does Kastra support?

Kastra provides one unified policy layer across AWS, Azure, GCP, and Oracle. It works across any cloud environment, partner networks, and on-premises infrastructure.