Annolid Documentation Portal

Run reproducible annotation and tracking workflows.

Annolid combines a desktop GUI, a composable CLI, and optional agent-assisted tools for real-world behavior analysis projects.

Start with a stable local setup, label and review a small video, then move into repeatable training, inference, behavior scoring, and automation.

Getting Started in 20 minutes Install Annolid Use .venv for development Follow core workflows Use behavior-agent analysis Run annolid-run from Annolid agents Track long videos with SAM3 Explore tutorials Enable memory subsystem

Choose Your Path

New Users Start with installation and one-line setup paths, then run your first GUI project.
Installation | One-Line Installer

Power Users Use annolid-run, MCP integrations, and automation-friendly CLI flows.
Workflows | Agent CLI | MCP | Reference

Maintainers Keep deployment, migration, and release workflows predictable and documented.
Deployment | Migration Plan

Core Areas

Setup

Pick the right environment path for your platform and development mode.

Open installation guide

Workflow Execution

Run GUI and CLI tasks with explicit command patterns that can be repeated.

Open workflows

Zone Analysis

Draw chamber layouts, reuse zone files, and export phase-aware summaries.

Open zone analysis reference

Agent CLI

Use the typed annolid_run tool for safe agent-driven CLI operations.

Open agent CLI guide

SAM3 Tracking

Use SAM3 and Annolid Bot for windowed tracking on long, drifting, or occluded videos.

Open SAM3 guide

Tutorials

Jump into focused guides for tracking, segmentation, and model operations.

Open tutorials

Memory System

Store reusable context, use scoped retrieval, and migrate legacy memory data.

Open memory docs

Agents and Security

Configure agents, isolate secrets, and validate local security posture.

Open security docs

Operations

Deploy docs and site assets, and keep release and migration flows in sync.

Open deployment guide

Product Snapshot

• Python package metadata supports >=3.10; the default GUI/core workflow is documented for Python 3.10-3.14.
• Primary entry points are annolid (GUI) and annolid-run (CLI/plugins).
• Memory subsystem includes GUI CRUD manager, structured settings profiles, and legacy-source migration tooling.
• Annolid Bot supports multimodal chat and optional provider integrations in the GUI.
• Docs are built with MkDocs Material in strict mode and published through GitHub Actions.