Back to directory

Autonomous Agent Framework (AAF)

A Python multi-agent platform built on a microservice, event-driven architecture, with a hybrid memory combining SQL, ChromaDB and a KuzuDB-based graph store, and Docker-in-Docker sandboxing that runs agent-written code in disposable, isolated containers.

Autonomous Agent Framework (AAF)

AAF is a multi-agent platform built as an asynchronous, event-driven system rather than a lightweight scripting wrapper. An event bus decouples a central reasoning process from its inputs, such as Telegram messages, system logs and background tasks, which arrive as a single priority queue.

Its memory combines SQL for fixed rules, ChromaDB for semantic similarity, and a KuzuDB-based graph store for non-linear associations between facts, so context is not limited to a short window. Code the agent writes is executed inside disposable, fully isolated Docker containers rather than on the host, and a built-in key manager rotates between multiple API keys with a round-robin algorithm when a rate limit is reached. A CLI manager lets several independent agents be deployed and orchestrated on one server, generating the needed Docker Compose configuration automatically.

It is written in Python, distributed as a Docker-packaged project, and documented primarily in Russian.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.