Engineering blog

Your agent runs in a disposable sandbox, but it can't live there. A breakdown of everything in the agent's world, and why it should be one forkable bucket.

Fork a customer's entire LangGraph agent state, replay their conversations through a prompt change, and judge it head-to-head before you ship to prod.

Kefka is a userspace shell sandbox in Go that gives every AI agent its own copy-on-write Tigris bucket fork plus Python, jq, and ripgrep via WebAssembly.

We embedded agent-shell on the Tigris homepage so you can ls, cat, and cd through your buckets like a normal filesystem. Here's how it works.

A self-updating knowledge base on Tigris, maintained overnight by an Anthropic SDK agent loop running through @tigrisdata/agent-shell. Atomic flush on success, discard on throw. The bucket either lands a clean run or stays byte-for-byte unchanged. ~$7 a month, presigned digest URL in Slack each morn…

Meet @tigrisdata/agent-shell: a virtual bash environment that gives AI agents persistent storage on Tigris buckets, with forks and snapshots for safety.

Storage-protected Mastra agents: a reference implementation. Isolated session forks, validate before promoting, contain compromise at the storage layer.

A new TypeScript SDK giving AI agents the storage primitives they need: forks, workspaces, checkpoints, and event-driven coordination on Tigris.

Bucket forking on Tigris gives every agent its own sandbox with no data duplication. One Lance file, two query engines, globally distributed reads.

Tigris now ships as a plugin for Claude Code, Cursor, and other AI coding agents. Five skills, a storage subagent, and security rules give your agent full Tigris fluency out of the box.

Agentuity built custom cloud infrastructure for agentic workloads. Tigris powers five storage layers, from agent filesystems to sandbox snapshots.

Prevent rogue agent behavior by giving every agent an instant, isolated bucket fork—no data duplication, safe concurrency, and easy rollback with snapshots.

Discover why OpenAI's gpt-oss model family is ideal for building reliable and safe AI agents, not just for developers.