mycode-sdk 0.9.4

Lightweight Python SDK for building AI agents.

mycode-sdk

Lightweight Python SDK for building AI agents.

Install

uv add mycode-sdk
# or
pip install mycode-sdk

Quick start

import asyncio

from mycode import Agent, bash_tool, read_tool


async def main() -> None:
    agent = Agent(
        model="claude-sonnet-4-6",
        api_key="YOUR_API_KEY",
        tools=[read_tool, bash_tool],
    )

    async for event in agent.achat("Read pyproject.toml and tell me the project name."):
        if event.type == "text":
            print(event.data["delta"], end="", flush=True)


asyncio.run(main())

Agent(...) infers the provider from the model id. No tools are registered unless you pass tools=[...].

For a simple synchronous call, use run():

result = agent.run("Read pyproject.toml and tell me the project name.")
print(result.text)

Multi-turn conversations

Call achat() or run() again on the same Agent — history accumulates automatically:

agent = Agent(model="claude-sonnet-4-6", api_key="...")

agent.run("What is 2 + 2?")
agent.run("Now multiply that by 10.")    # remembers the earlier answer

Attachments

Pass attachments to achat() or run() to add files alongside the prompt:

from mycode import Attachment

agent.run("Describe these.", attachments=["diagram.png", "report.pdf", "notes.txt"])

# Or build them explicitly — raw bytes and inline text never touch the disk:
agent.run(
    "Review.",
    attachments=[
        Attachment.path("diagram.png"),
        Attachment.bytes(png_data, media_type="image/png"),
        Attachment.text("TODO: ship it", name="note.md"),
    ],
)

Images support image/png, image/jpeg, image/gif, image/webp; documents support application/pdf. Sending an image or PDF to a model that lacks that capability yields an error event; a bad path or unsupported type raises ValueError before the model is called.

Saving sessions

Pass session_dir to persist the conversation to disk. Each session lives in a subdirectory named by session_id:

from pathlib import Path

agent = Agent(
    model="claude-sonnet-4-6",
    api_key="...",
    session_dir=Path("./chats"),
    session_id="my-chat",
)

Construct another Agent with the same (session_dir, session_id) later to resume the conversation — the history is loaded automatically.

Built-in tools

from mycode import read_tool, write_tool, edit_tool, bash_tool

Four tools for reading, writing, editing files, and running shell commands. Opt in by passing them to tools=[...].

Custom tools

Decorate a typed function with @tool:

from mycode import Agent, tool


@tool
def greet(name: str) -> str:
    """Return a friendly greeting.

    Args:
        name: Person name.
    """

    return f"hello, {name}"


agent = Agent(
    model="claude-sonnet-4-6",
    api_key="...",
    tools=[greet],
)

To call a built-in tool from inside your own tool, type the first parameter as ToolContext:

from mycode import ToolContext, tool


@tool
def summarize_file(ctx: ToolContext, path: str) -> str:
    """Return the first line of a text file."""

    result = ctx.read(path)
    return result.output.splitlines()[0] if result.output else ""

Tool hooks

Inspect or replace tool calls before they run. Return None from before_tool to let the tool execute, or a ToolExecutionResult to skip it:

from mycode import Agent, Hooks, ToolExecutionResult, bash_tool

hooks = Hooks()


@hooks.before_tool
async def block_rm(ctx):
    cmd = str(ctx.tool_input.get("command") or "")
    if ctx.tool_name == "bash" and "rm -rf" in cmd:
        return ToolExecutionResult(output="error: blocked", is_error=True)
    return None


agent = Agent(
    model="claude-sonnet-4-6",
    api_key="...",
    tools=[bash_tool],
    hooks=hooks,
)

@hooks.after_tool runs after the tool and can replace the result (audit, redact, etc.).

See docs/sdk.md for the event stream, cancellation, sessions, and the full Agent / @tool reference.