glia¶
A small library for building LLM agents. Every model call, tool call and state change is a plain object that can be logged, saved and replayed. The agent loop is a single file.
The name comes from glial cells, which support and connect neurons.
Why¶
Agent frameworks tend to hide the control flow, which makes them hard to debug. glia keeps the loop small enough to read. Tools, structured outputs, streaming, parallel tool execution, context compaction, checkpoints, guardrails, a human-in-the-loop approval gate, subagents and evaluations written as tests are optional modules on top of it.
Install¶
pip install glia-agents # core — no dependencies
pip install "glia-agents[anthropic]" # + the Claude provider
The distribution is glia-agents; the import stays import glia.
30 seconds¶
import asyncio
from glia import Agent, tool
from glia.providers import ClaudeLLM
@tool
async def get_weather(city: str) -> str:
"""Get the current weather for a city."""
return {"Paris": "18°C, cloudy"}.get(city, "unknown")
async def main():
agent = Agent(ClaudeLLM(), tools=[get_weather], system="Be concise.")
result = await agent.run("What's the weather in Paris?")
print(result.output) # the answer
print(result.usage) # what it cost
asyncio.run(main())
No API key? Every example runs offline with the deterministic EchoLLM
provider — same code, no network.
Next¶
- Getting started — install, first agent, streaming, offline testing
- Guide — every primitive, with code
- Architecture — how the whole thing works
- Strategy — market analysis and positioning