claude-agent-sdk-python
claude-agent-sdk-python copied to clipboard
anthropics
## Summary Adds automatic accumulation of streaming text/thinking deltas into `TextBlock` and `ThinkingBlock` objects, making it much easier to build real-time UIs that display text as it streams from the...
### Summary Subagent (Task tool) messages are saved to `history.jsonl` with `isSidechain: true` but occasionally are not yielded through the `receive_response()` async generator. This causes assistant responses from subagents to...
Add support for limiting API costs using the max_budget_usd option, mirroring the TypeScript SDK functionality. When the budget is exceeded, query execution stops and returns a result with subtype 'error_max_budget_usd'....
As per the official mcp SDK's mcp.types: `ContentBlock = TextContent | ImageContent | AudioContent | ResourceLink | EmbeddedResource` but the claude agent sdk (version 0.1.4) currently does this in the...
``` options = ClaudeAgentOptions( mcp_servers={ "chrome-devtools": { "type": "stdio", "command": "npx", "args": [ "-y", "chrome-devtools-mcp@latest", "--isolated" ] } }, ) ``` I'm seeing that any prompt which launches a browser,...
Sometimes, I want to do something outside the current directory, such as, `ls /var/log`. But I didn't find the way to approve it. But in CLI, I only need to...
Hello! I'm building my own workflows on top of Claude Code using this SDK. Some of these workflows require a human in the loop. Everything works Ok except for permission...
## Issue [The v0.1.0 migration guide](https://docs.claude.com/ja/docs/claude-code/sdk/migration-guide) states: > "The SDK no longer uses Claude Code's system prompt by default." However, this breaking change is **not implemented**. ## Current Implementation In...
While the SDK supports streaming for various message blocks (e.g., tool usage/result, system message, result message, text message etc.), `TextBlock` content is only delivered once it's fully available. This delays...
