Plugins¶

letscode's plugin system is built on pluggy. A plugin is a Python package that registers tools, slash commands, skills, frontends, or lifecycle hooks via standard Python entry points. letscode discovers plugins at startup, calls their registration hooks, and lets the lifecycle hooks intercede during agent turns.

Authoring guide

Step-by-step: what to register, what hooks fire when, how to distribute via PyPI.
Reference plugin: letscode-memory

The canonical worked example — cross-session memory in ~150 lines.
Extension model contract

Stable surface, versioning rule, the ten lifecycle hooks, design rationale.

What you can register¶

Five extension types:

Type	Hook	What it adds
Tool	`letscode_register_tools`	LLM-callable tools (e.g. `read`, `write`).
Command	`letscode_register_commands`	Slash commands (e.g. `/help`, `/skills`).
Skill	`letscode_register_skills`	Skill source dirs or inline skills.
Frontend	`letscode_register_frontends`	Named frontend factories.
Lifecycle hook	`@hookimpl` on any of the lifecycle hookspecs	Cross-cutting behavior.

The first four are named things added to a registry. The fifth is behavior layered on the agent loop. See Extension model for the full hook inventory.

TL;DR — your first plugin¶

mypkg/plugin.py

import pluggy
from pydantic import BaseModel

from letscode.agent.tools import ToolContext, ToolResult, tool
from letscode.llm.types import TextPart

hookimpl = pluggy.HookimplMarker("letscode")


class _GreetParams(BaseModel):
    name: str


@tool(name="greet", description="Say hello to someone.")
async def _greet(params: _GreetParams, ctx: ToolContext) -> ToolResult:
    del ctx
    return ToolResult(content=[TextPart(text=f"Hello, {params.name}!")])


@hookimpl
def letscode_register_tools(registry):
    registry.add(_greet)

pyproject.toml

[project.entry-points.letscode]
my_plugin = "mypkg.plugin"

pip install -e .          # or: pip install mypkg
letscode "say hi to alice"

The LLM can now call the greet tool. That's the whole story for tools.

For everything else — commands, skills, frontends, lifecycle hooks, message types, packaging, testing — see the authoring guide.