Skip to content
AnyCapAnyCap Docs

Capabilities Overview

AnyCap groups capabilities around four jobs: create something, understand something, find current information, and deliver a result.

Understand media

Analyze images, video, and audio without managing a separate vision or transcription service.

Use the web

Find structured results, synthesize cited answers, and convert known pages into Markdown.

Deliver and collaborate

Collect visual feedback, share files, publish pages, and hand projects between agents.

Generation workflow

Section titled “Generation workflow”

Image, video, music, and audio use a live model catalog. Do not assume that a model, mode, or parameter remains available.

Terminal window
# 1. Discover models
anycap <capability> models


# 2. Inspect the selected contract
anycap <capability> models <model-id> schema --mode <mode>


# 3. Generate and name the output
anycap <capability> generate \
  --model <model-id> \
  --mode <mode> \
  --prompt "..." \
  -o descriptive-output.ext

Local files

Section titled “Local files”

Compatible commands accept a local path directly and upload it internally. Use --file for understanding actions and the model’s documented --param field for generation. Drive is for sharing results with humans, not for moving files between AnyCap commands.

JSON-first output

Section titled “JSON-first output”

Every non-MCP command writes one final JSON document to stdout. Generation responses include local output paths, model information, credit usage when available, and diagnostic IDs.