edgecrab 0.12.1

EdgeCrab AI Agent SDK — build autonomous agents with Python

EdgeCrab Python SDK

WHY

Use this SDK when you want Python productivity with the Rust runtime doing the heavy lifting.

This is the canonical EdgeCrab Python SDK publication path.

It is designed for:

• async or sync agent workflows
• local Ollama validation before paid-provider rollout
• session-aware applications
• memory-backed assistants and research tools

WHAT

Core capabilities:

• native Rust-backed runtime via PyO3
• sync and async agents
• streaming callbacks
• persistent sessions and memory
• profile-aware configuration via Config.load_profile()

Architecture

Python app
   |
   v
PyO3 bindings
   |
   v
EdgeCrab Rust runtime
   |
   +--> sessions
   +--> memory
   +--> tools
   |
   v
LLM provider or local Ollama

HOW

Quick start

python3 -m pip install -U edgecrab

from edgecrab import Agent

agent = Agent(model="copilot/gpt-5-mini")
reply = agent.chat_sync("Hello")
print(reply)

Development and examples

See the full examples guide in examples/README.md.

cd sdks/python
python3 -m pip install -U maturin
maturin develop
make e2e

SDK Tutorials

All tutorials ship with working Python examples in examples/.

Tutorial	Example	Model
1. Cost-Aware Code Review	examples/cost_aware_review.py	openai/gpt-5 + copilot/gpt-5-mini
2. Parallel Research Pipeline	examples/parallel_research.py	copilot/gpt-5-mini
3. Multi-Agent Pipeline	examples/multi_agent_pipeline.py	openai/gpt-4o + copilot/gpt-5-mini
4. Session-Aware Support Bot	examples/session_aware_support.py	copilot/gpt-5-mini
5. Safe SQL Agent	examples/safe_sql_agent.py	copilot/gpt-5-mini