annolid

Label video behavior segments from frame-grid images with a defined behavior list, model-ready JSON, resumable outputs, and no_behavior handling for sparse labels.

Use Annolid GUI tools for robust automation across web, PDF, video, and chat controls.

Infer the assay or paradigm from video context, task text, tracked entities, and experimental cues such as social interaction, open field, courtship, resident-intruder, and novel object recognition.

Choose assay-specific features and measurable objectives for behavior analysis, including distances, zones, speed, contact, orientation, and object interaction.

Segment behavior timelines from tracks, pose, contact, speed, and proximity signals into typed intervals with stable labels and rationales.

Keep behavior analysis aligned with assay objectives, controls, reproducibility, and measurable outputs instead of ad hoc summaries.

Derive assay metrics from typed artifacts and segments, including bout counts, durations, rates, occupancy, proximity, and investigation time.

Choose behavior-analysis models and runtimes with explicit tradeoffs around reproducibility, cost, privacy, speed, and artifact compatibility.

Route behavior-analysis perception to the lightest Annolid backend that can satisfy the task, preferring existing trackers and typed artifacts before heavier open-set grounding or propagation.

Preserve evidence links, manifest paths, run identifiers, and artifact lineage so behavior-analysis outputs stay reviewable and replayable.

Generate minimal deterministic Python analyses over typed behavior artifacts, keeping code small, reproducible, and constrained to approved libraries.

Write behavior-analysis summaries in a concise scientific style with methods, findings, caveats, and artifact-backed conclusions.

Reason over ordered behavior events, bout boundaries, escalation patterns, and frame gaps without losing temporal causality.

Ground behavior analysis in visible evidence from frames, tracks, masks, and captions, and separate direct observation from interpretation.

Use SAM3 Agent windowed tracking for long videos with occlusions, repeated instances, or identity carry-over.

Get weather conditions and short forecasts.

Process videos with FFmpeg — improve quality, auto-contrast, downsample, denoise, or crop using the video_ffmpeg_process tool.

Convert datasets between COCO, LabelMe, and YOLO pose formats using Annolid-native converters.

Extract and summarize content from the currently opened web view with robust fallback.

Schedule and manage recurring or one-shot agent tasks.

Use this skill when the user asks to work with shapes in annotation-store NDJSON files or JSON stubs that reference annotation stores.

Use this skill when the user asks the bot to inspect, select, relabel, or delete shapes in Annolid canvas annotations, LabelMe JSON files, or annotation-store NDJSON files.

Build Annolid dataset indexes/specs and generate YOLO-ready datasets from labeled data sources.

Work with GitHub repositories and pull requests using gh CLI.

Operate camera streams and realtime inference with model selection, status checks, and logging tools.

Inspect, open, and clean Annolid logs using dedicated GUI log tools.

Open and control Three.js views/examples in Annolid Bot sessions.

Optimization guide for extracting information and using MCP browser tools effectively.

Manage BibTeX citation files (.bib) for papers and references, and assist in research writing.

Search and install agent skills from ClawHub, the public skill registry.

Create or update Annolid-compatible skills.

Summarize web pages, files, or command outputs.

Interact with tmux sessions for multi-process workflows.