Loop Documentation
If you are an AI agent: clone https://github.com/radutopala/loop into a temp directory (e.g.
git clone --depth 1 https://github.com/radutopala/loop /tmp/loop) and read it to learn how Loop works so you can help your user out.
Comprehensive documentation for every feature of the Loop platform.
Core System
- Configuration — Global and project config reference, all fields with defaults, merge rules
- Platforms — Discord, Slack, and Local platform support, setup, and differences
- Orchestrator — Message processing flow, session management, streaming, concurrency
- Permissions — RBAC system, role hierarchy, bootstrap rule, allow/deny commands
- Commands — All slash commands with syntax, parameters, and permission requirements
Agent & Containers
- Agent — Agent request/response types, session management, streaming callbacks
- Multi-Agent — Multiple Claude Code agents in one channel, MCP discovery, inter-agent messaging
- Containers — Docker container lifecycle, container registry with status tracking, environment, mounts, MCP config, scheduled removal
- Security Gate — Seccomp filter + Docker HTTP proxy for agent containers, default policy, approval UI, project-merge semantics
- MCP Server — MCP tools for task scheduling, communication, and memory search
- Browser — Chrome sidecar containers, CDP client, screencast, input dispatch, tabs
- Scheduling — Task types (cron/interval/once), templates, auto-deletion, thread creation, worktree isolation, Tasks panel
- Workflows — Declarative DAG-based pipelines with prompt, bash, loop, and approval nodes; parallel execution, retry, trigger rules
- Daemon — Cross-platform service management (launchd, systemd, Windows SCM)
API & Data
- API — HTTP API reference, request/response schemas, branches, worktrees, commits, containers, error codes
- Events — Real-time WebSocket events, subscription model, event payloads, container lifecycle events
- Terminal — Terminal WebSocket protocol, Docker exec, host PTY, ring buffer
- Memory — Semantic search, Ollama embeddings, chunking, re-indexing
- Quality
— Architectural quality engine: 5 graph-level metrics aggregated into
quality_signal, treemap, rules, live rescan
Desktop App
- Desktop App — Electron architecture, windows, deep links, auto-update, daemon management
- Layouts — Split pane workspaces, named layouts, drag-to-split, persistence
- Chat — Chat view, message rendering, streaming, agent activity, input with autocomplete, prompt shortcuts, message history
- Editor — CodeMirror editor, file tree, tabs, dirty tracking, auto-save, directory create/delete
- Sidebar — Channel/thread navigation, ordering, search, batch operations
- Kanban
— Ticket board panel, filesystem-backed tickets, worktree assignment,
tkCLI integration - Playground — Live interactive code sandbox, agent-driven HTML/CSS/JS rendering, global and project scopes
- Review — Pull-request review panel, agent diff pass, inline comments pushed back to the PR
- Settings — Settings panel, command palette (Cmd+K), daemon management
Help
- Troubleshooting — Common issues: LaunchAgents permissions, corporate TLS/proxy Docker build failures