A developer open-sourced 'brain' (github.com/aexhq/brain), a minimal, extensible agent runtime inspired by Anthropic's managed agents engineering post and pi.dev. It targets production agent use cases the author found too rigid in LangChain and Mastra. For practitioners, it offers a lightweight alternative with sub-second cold starts and a small compiled footprint.
Brain is an open-source agent platform positioned as a smaller, more flexible alternative to LangChain and Mastra, aimed at production agent use cases. It draws on Anthropic's managed agents engineering post and pi.dev. It targets developers who find existing agent frameworks too complicated or rigid.
Brain is an open-source agent platform positioned as a smaller, more flexible alternative to LangChain and Mastra, aimed at production agent use cases. It draws on Anthropic's managed agents engineering post and pi.dev. It targets developers who find existing agent frameworks too complicated or rigid.
Phoson is a framework-free, open-source Python runtime for running AI agents, paired with a terminal CLI. It uses LLMs as the agent backbone while keeping the harness small and replaceable, avoiding heavyweight frameworks like LangChain. It targets developers who find agent frameworks more complicated than the agents themselves.
Phoson is a framework-free, open-source Python runtime for running AI agents, paired with a terminal CLI. It uses LLMs as the agent backbone while keeping the harness small and replaceable, avoiding heavyweight frameworks like LangChain. It targets developers who find agent frameworks more complicated than the agents themselves.
Athena is an open-source Python experiment exploring what infrastructure around an LLM is needed to make it useful beyond chat. It runs locally via Ollama with optional cloud backends and combines an agent loop, tools, memory, document ingestion, web access, and permission guardrails. It targets developers who want a sandbox for building local-first agents rather than a production platform.
Athena is an open-source Python experiment exploring what infrastructure around an LLM is needed to make it useful beyond chat. It runs locally via Ollama with optional cloud backends and combines an agent loop, tools, memory, document ingestion, web access, and permission guardrails. It targets developers who want a sandbox for building local-first agents rather than a production platform.
A developer launched KISS on GitHub, a performant agent harness written in Rust and modeled on the Pi harness, created after frustration with Pi's performance. It targets practitioners who want an embeddable, multi-language agent runtime.
A developer launched KISS on GitHub, a performant agent harness written in Rust and modeled on the Pi harness, created after frustration with Pi's performance. It targets practitioners who want an embeddable, multi-language agent runtime.
SynapsCLI is a lightweight agent runtime written in Rust that orchestrates a swarm of workers while minimizing LLM costs and resource usage. It boots in 20ms, supports any model via OAuth, API keys, or local models, and is extensible through hooks, plugins, and skills.
SynapsCLI is a lightweight agent runtime written in Rust that orchestrates a swarm of workers while minimizing LLM costs and resource usage. It boots in 20ms, supports any model via OAuth, API keys, or local models, and is extensible through hooks, plugins, and skills.