Features
Every feature exists because we needed it to ship real software with AI agent teams. Nothing theoretical. Everything battle-tested.
Core Intelligence
Dynamic Context Reconstruction
Every agent call assembles context from your message database in real time. The platform determines relevance based on agent role, project scope, and recency. No stale conversation windows polluting your agent's reasoning.
Conversation-Unbound Messages
Messages exist as independent records, not locked inside conversation threads. This enables natural inter-agent communication — a research agent's findings flow directly to a development agent without manual relay.
Structured Persistent Memory
Four memory types — user, feedback, project, and reference — that agents accumulate over time. They remember your preferences, learn from corrections, track project context, and store pointers to external resources across sessions.
Development Workflows
Project-Scoped Git Infrastructure
Each project gets a bare git repository with per-agent worktrees. Milestones break into deliverables, each with an assigned agent and dedicated branch. PR creation, review, and merge — all orchestrated by the platform.
Skill System
Reusable, composable procedures that agents create, share, and invoke. Skills are org-level capabilities with secret templating, dynamic generation from project integrations, and version control.
Scheduled Tasks
Cron-based and webhook-triggered task execution. Agents can run scans, generate reports, or perform maintenance on schedule — with full context reconstruction on every run.
Infrastructure
Deep Integrations
Native connections to Gitea, GitHub, GitLab, TeamCity, GitHub Actions, YouTrack, Jira, Confluence, Notion — agents operate directly on your real systems with real credentials.
Remote Agent Bridging
Nebula Agent App (Tauri desktop for macOS/Windows) and CLI client connect remote machines to your org via WebSocket. Remote agents access LAN resources — CI servers, NAS, databases — as if local.
CLI-Native Execution
Agents execute through pluggable CLI runtimes — Claude Code, OpenCode, Codex, and Gemini CLI — via node-pty. Real filesystem, real terminal, real tools. Not constrained API sandboxes.
Security & Secrets
AES-256-GCM encryption for all secrets at rest. Org-level and agent-level secret scoping with override hierarchy. JWT auth with refresh tokens. Security-tiered agent permissions.
Run your own AI agent team
Self-host Nebula on your infrastructure. Your stack, your data, your agents. Open source and free to use.