Skip to content

RELEASE_NOTES_v3.1.0.md

Latest commit

 

History

History
91 lines (66 loc) · 3.57 KB

RELEASE_NOTES_v3.1.0.md

File metadata and controls

91 lines (66 loc) · 3.57 KB
Raw

Operon v3.1.0 — Initial Public Beta

Operon is an advanced autonomous AI Terminal Cockpit — a fully agentic, multi-provider terminal AI with 185+ tools, persistent memory, browser & desktop automation, a multi-agent mesh, and a self-improvement loop.

This is the first public release. It's feature-rich and well-tested (2,750+ tests passing), runs on macOS/Linux/Windows, and works fully offline with local models via Ollama.

Install in one command

git clone https://github.com/OperonAgent/Operon.git
cd Operon
./install.sh            # macOS/Linux   (Windows: install.ps1)
operon

The installer provisions everything — Python deps and the Playwright Chromium browser binary (the ~120 MB download a plain pip install skips). Pre-built Operon.app / operon.exe / Linux binaries are attached below.

Highlights

  • 185+ tools — files, shell, git, HTTP, databases, PDF, vision, image gen, TTS, video gen
  • 8+ AI providers — OpenAI, Anthropic, OpenRouter, Ollama, LM Studio, Jan, and more
  • Persistent memory — FTS5 keyword + LanceDB semantic vectors + Obsidian vault sync, unified behind one facade
  • Browser & computer use — Playwright automation + pyautogui desktop control, with runtime self-heal
  • SWE agent — automated issue → fix → test → PR loop, now with LSP-style static analysis
  • Multi-agent mesh — parallel/pipeline delegation with named specialist roles
  • Concurrent tool dispatch — independent tool calls run in parallel
  • Voice pipeline — speech-to-text, text-to-speech, real-time streaming STT
  • Self-improvement — synthesizes new Python tools from conversation trajectories
  • Security — email_send is structurally un-callable by the model; OSV.dev CVE dependency scanning in /doctor
  • 60+ slash commands — /kanban, /checkpoint, /mesh, /swe, /voice, /doctor, …
  • Claude Code-style TUI — persistent input bar, token counter, model HUD
  • Plugin SDK — manifest + tools, with 5 ready-to-use example plugins

Since the project's internal builds

  • One-command installer + dependency bootstrapper with browser self-heal
  • Supply-chain CVE auditing (core/dep_audit.py) wired into /doctor
  • Unified memory facade over FTS5 + vector + Obsidian
  • 5 seed example plugins (11 tools) demonstrating the SDK
  • Video generation tool (Replicate / Luma)
  • Static analysis in the SWE loop
  • Began breaking up the monolithic main.py into a cmd_handlers/ package
  • 120+ new tests, including real agent-loop integration tests
  • Full repo hygiene: LICENSE, README, CHANGELOG, CONTRIBUTING, SECURITY, ROADMAP

Fixes

  • Repaired 6 broken slash commands (/vector, /desktop, /synth, /checkpoint, /kanban list, /usage, /knowledge)
  • Restored the TUI _extra_status back-compat alias
  • Hardened /doctor against import-time errors in optional deps

Known limitations (tracked in ROADMAP.md)

  • main.py modularization is in progress (read-only commands extracted; global-state commands pending)
  • Messaging channels (Discord/Telegram/Slack) are functional but not yet at production depth
  • Voice streaming to cloud realtime APIs is partial

Requirements

  • Python 3.9+ (3.11+ recommended), macOS / Linux / Windows
  • Optional: Ollama for fully-offline local models (no API key)

Docs

  • Setup Guide: Operon_Setup_Guide.pdf
  • Full reference: Operon_Documentation.pdf
  • Comparison: Operon_Comparison.pdf
  • Build them: make docs

Full Changelog: https://github.com/OperonAgent/Operon/blob/main/CHANGELOG.md