meta:media

Store and search multimodal media with ChromaDB hybrid semantic and metadata retrieval.

264|11|Updated Apr 25, 2026
One-click install
npx skills add https://github.com/coco-research/coco --skill meta-media-coco-research
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: meta:media
Source: https://github.com/coco-research/coco/tree/main/skills/media-memory
Command: npx skills add https://github.com/coco-research/coco --skill meta-media-coco-research

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill turns scattered media into a searchable, persistent memory so you can save, find, and reuse images, video, audio, documents, and generated assets without losing context.

Core Features & Use Cases

  • Multimodal Ingestion: Stores user-supplied or generated media with rich descriptions, tags, source tracking, and extracted text or transcripts.
  • Hybrid Retrieval: Searches by semantic meaning and metadata across screenshots, diagrams, mockups, recordings, and other past assets.
  • Proactive Recall: Surfaces relevant media when an ongoing conversation matches something already stored in memory.
  • Use Case: A teammate asks for the architecture diagram from last week, and the Skill locates the exact file, shows its metadata, and makes it easy to reopen or reuse.

Quick Start

Save the attached media into memory with a rich description, relevant tags, the correct source, and any extracted text.

Frequently Asked Questions about meta:media

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search and retrieve past media assets like screenshots and recordings?

You can search and retrieve past media assets using hybrid semantic search combined with metadata filtering. This approach matches conversational meaning against stored descriptions, tags, and extracted text to instantly locate specific screenshots, recordings, or documents.

What is multimodal media memory and when do I need it for ongoing conversations?

Multimodal media memory stores images, video, audio, and documents with rich descriptions, tags, and extracted transcripts. It solves the problem of scattered assets by providing persistent tracking, enabling you to find and reuse files without losing context.

How do I store generated images and diagrams with rich metadata for later recall?

You store generated images and diagrams by ingesting them with detailed descriptions, relevant tags, source tracking, and any extracted text. This rich metadata capture enables accurate hybrid retrieval and proactive recall during future conversations.

Can I use ChromaDB for local embedding generation to search media memory?

Yes, ChromaDB enables local embedding generation to search media memory effectively. It generates semantic embeddings locally from your stored images, video, audio, and documents, allowing hybrid semantic-plus-metadata retrieval without relying on external embedding APIs.

Does hybrid retrieval work for finding specific architecture diagrams and mockups?

Yes, hybrid retrieval works for finding specific architecture diagrams and mockups by searching across both semantic meaning and metadata. It indexes extracted text and descriptions to locate exact files and display their metadata for easy reopening.

How do I store generated images and diagrams with rich metadata for later recall?

You store generated images and diagrams by ingesting them with detailed descriptions, relevant tags, source tracking, and any extracted text. This rich metadata capture enables accurate hybrid retrieval and proactive recall during future conversations.