Skip to content

Roadmap

Current Release: v0.1.4 (Alpha)

Completed

  • 17 built-in tools (file I/O, shell, web, planning, memory, identity, skills)
  • Multi-provider LLM support (OpenAI-compatible, Ollama, vLLM)
  • MCP server integration
  • Skills system with LLM-driven activation
  • LangGraph-powered workflow engine
  • Layered permission system (default, auto-edit, yolo)
  • Memory and identity profiles with multi-agent isolation
  • Session management with LLM-based compression
  • Cost tracking per provider and model
  • Gateway daemon with Telegram, Discord, Slack, Feishu adapters
  • Docker support (GHCR)
  • Embeddable AgentCore Python API

In Progress

  • Documentation improvements
  • Media generation support (image, audio)

Planned

  • Deep Searchcliver deep-search "question" multi-step research workflow
  • Multi-Model Orchestration — route different capabilities to different models
  • Remote Workflow Definitions — fetch workflows from remote sources
  • Plugin System — installable capability packages

Contributing

See CONTRIBUTING.md for how to get involved.

Have an idea? Open a feature request or start a discussion.