Kastra
Control what your AI is allowed to do. Sub-millisecond authorization for every prompt, tool call, shell command, and API request.
Quick answer
What is Kastra?
At a glance
- Category
- AI Agents & Automation
- Pricing
- Not disclosed
- Starting price
- Not disclosed
- Free tier
- No
- Platforms
- Not disclosed
- Launched
- Not disclosed
- Company
- Not disclosed
About Kastra
Control what your AI is allowed to do. Sub-millisecond authorization for every prompt, tool call, shell command, and API request. Before an action runs, Kastra evaluates attribute-based policy in under one millisecond and returns allow, deny, redact, or hold. Recorded decisions form a cryptographically chained audit trail so operators, auditors, and agents can explain why something was permitted. Use the Kastra HTTP API from production agents, install Kastra Edge on developer machines to govern Claude Code, Cursor, and Codex, or connect an MCP client to read governance data.
From the community
User reviews
No user reviews yet. Be the first to share your experience with Kastra.
Alternatives
Similar to Kastra
Bol.ai
Stop retyping Bills of Lading. Upload a BOL as PDF, scan or photo and get structured data in seconds, parties, ports, containers, weights. JSON, CSV & API.
LockIn MCP
Tell your AI to block distractions. Then get back to work. Distraction blocker for macOS, Linux & Windows. Pro From $2.99/month. Works with Poke, Claude & ChatGPT.
NodePad
Agent on canvas. NodePad lays your AI conversations out on a canvas, so you try more ideas, keep the answers that hold up, and never lose the thread that mattered.
Aidress
are utilities designed to be called by a model: their inputs are designed to be generated by models, and their outputs are designed to be passed back to models.
Outside Agent
Build AI agents that interact with the outside world.
Reduck
Ready-made automation scripts for LinkedIn, Twitter, Reddit and more, or write your own.
Techavy is reader supported. When you buy through links on our site we may earn an affiliate commission. This never changes our ratings or what we recommend. See our methodology.