Security advisory — Malicious litellm versions 1.82.7 and 1.82.8 were removed from PyPI (potential API key exfiltration). Uninstall them, rotate exposed credentials, and upgrade to a safe release (e.g. 1.82.9+). Run pip show litellm to verify. PyPI · README

Roadmap

AgenticX development roadmap.

Roadmap

This map is a module index, not a promise that every adapter is production-complete. Studio still does not send every operation through one SafetyLayer. Cluster / HA Runtime and a default Edge Agent path are planned, not the current Near or Enterprise default.

Diagram: shipped core vs planned M12, M18, and cluster runtime.

Completed Modules

ModuleDescription
M1Core abstractions — Agent, Task, Tool, LLM Provider interfaces
M2LLM Provider layer — 15+ providers with unified interface
M3Tool System — @tool decorator, MCP Hub, Remote Tools v2, OpenAPI toolset
M4Memory System — Hierarchical memory, Mem0 integration, workspace memory
M5Agent Core — Think-act loop, 12-Factor methodology, self-repair, event-driven
M6Orchestration Engine — Graph-based workflows, conditional routing, parallel execution
M7Task Validation — Pydantic output parsing, auto-repair, guiderails
M8A2A Protocol — Inter-agent communication, AgentCard, skill-as-tool
M9Observability — Callbacks, metrics, Prometheus/OTel, trajectory analysis
M10Storage Layer — KV/Vector/Graph/Object with unified manager
M11Safety Layer — Injection detection, policy engine, audit logging, sandbox
M13Knowledge & GraphRAG — Document pipeline, chunkers, graph builders, retrievers
M14Avatar & Group Chat — Registry, routing strategies, delegation
M15Studio Runtime — Session manager, Meta-Agent, team manager
M16GUI Agent — Desktop automation, VLM reflection
M17Near Desktop — Electron + React, multi-pane workspace
v0.5Embeddable ReActAgent, unified offload, skill self-evolution, multi-brain knowledge, long-horizon coding

Planned

ModuleDescription
M12Agent Evolution: architecture search, knowledge distillation, adaptive planning. Skill self-evolution already shipped as a narrower loop.
M18Core / Studio multi-tenancy and RBAC on the Studio main path. Enterprise IAM already lives on a separate product line.

Cluster / HA Agent Runtime and a default Edge Agent path are also planned. They are not the current Near or Enterprise default.

Community Requests

Have a feature request? Open an issue →

Contribution

Want to help build these features? See CONTRIBUTING.md on GitHub.