Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 38 additions & 1 deletion integrations/agent-sdks.mdx
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
---
title: SDKs and frameworks
description: "Run agents in Sprites with adapters for Hugging Face OpenEnv, Google ADK, OpenAI Agents SDK, OpenRouter Agent SDK and TanStack AI, with runnable examples."
description: "Run agents in Sprites with adapters for Hugging Face OpenEnv, Google ADK, OpenAI Agents SDK, OpenRouter Agent SDK, Pydantic AI and TanStack AI, with runnable examples."
---

These adapters let an application use Sprites as the execution environment for an agent.
Expand Down Expand Up @@ -200,6 +200,43 @@ A successful run prints the agent's report of the sandbox's working directory.

`workspace.close()` releases adapter-local resources but preserves the Sprite. `workspace.destroy()` is the explicit infrastructure deletion operation.

## Pydantic AI

Pydantic AI Harness includes a `SpritesSandbox` capability that runs an agent's shell commands and file edits in a persistent Sprite. It requires Python 3.11 or newer.

```sh
pip install "pydantic-ai-harness[sprites,coder,anthropic]"
export SPRITE_TOKEN='...'
export ANTHROPIC_API_KEY='...'
```

The `coder` extra is needed for `Coder()`. The `anthropic` extra is only for the example model. Swap it for your model provider's extra.

Save this as `sprite_agent.py`. It gives the agent the harness's coding tools and asks it to report its working directory:

```python
from pydantic_ai import Agent
from pydantic_ai_harness.coder import Coder
from pydantic_ai_harness.sprites_sandbox import SpritesSandbox

agent = Agent('anthropic:claude-opus-5-5', capabilities=[SpritesSandbox(), Coder()])
result = agent.run_sync('Run pwd in the sandbox and report the working directory.')
print(result.output)
print(result.workspace.ref)
```

Run it with:

```sh
python sprite_agent.py
```

A successful run prints the agent's report of the sandbox's working directory, then the Sprite's workspace ref.

Pydantic AI never deletes the Sprite. It sleeps when idle and keeps its files until you delete it. Store `result.workspace.ref` to reattach later with `workspace=ref`, or delete the Sprite with `await SpritesSandbox().destroy(ref)`.

See the [Pydantic AI Sprites sandbox documentation](https://pydantic.dev/docs/ai/harness/sprites-sandbox/) for reattaching, cleanup after failed runs, durable execution, and configuration options.

## TanStack AI

TanStack AI provides a published Sprites sandbox provider with a durable filesystem, resume-by-ID, in-place checkpoints, and one proxied HTTP port. It requires Node.js 22.4 or newer.
Expand Down
Loading