UNPKG

aiwg

Version:

Cognitive architecture for AI-augmented software development with structured memory, ensemble validation, and closed-loop correction. FAIR-aligned artifacts, 84% cost reduction via human-in-the-loop, standards adopted by 100+ organizations.

145 lines (111 loc) 6.07 kB
# v2026.2.7 - "Media Curator" Release **Released**: February 14, 2026 This release introduces the **Media Curator framework** — a complete AIWG framework for building and maintaining intelligent media archives. It provides 6 specialized agents, 9 commands, and 9 skills covering the full pipeline from discography analysis through multi-platform export. ## Highlights | What Changed | Why You Care | |--------------|--------------| | **New media-curator framework** | Complete framework for AI-powered media archive management | | **6 specialized agents** | Discography analysis, source discovery, acquisition, quality assessment, metadata curation, completeness tracking | | **9 commands + 9 skills** | Full pipeline from `/analyze-artist` through `/export` | | **Field-tested patterns** | GAP-NOTE.md, opustags, production-context classification — proven on 94GB prototype | | **5 export profiles** | Plex, Jellyfin, MPD, mobile, archival — out of the box | ## The Framework The Media Curator framework (`agentic/code/frameworks/media-curator/`) follows the standard AIWG pattern: specialized agents orchestrated via commands, with skills providing reusable capabilities. ### Architecture ``` media-curator/ ├── agents/ # 6 specialized agents │ ├── discography-analyst # Era/project identification │ ├── source-discoverer # Finding content across platforms │ ├── acquisition-manager # Download orchestration │ ├── quality-assessor # Accept/reject criteria │ ├── metadata-curator # Tagging and organization │ └── completeness-tracker # Gap analysis ├── commands/ # 9 slash commands │ ├── analyze-artist # Discography research │ ├── find-sources # Source discovery │ ├── acquire # Download management │ ├── tag-collection # Metadata application │ ├── check-completeness # Gap analysis │ ├── assemble # Narrative assembly │ ├── curate # Main entry point │ ├── export # Multi-platform output │ └── verify-archive # Integrity verification ├── skills/ # 9 reusable skills │ ├── youtube-acquisition # yt-dlp patterns │ ├── archive-acquisition # Internet Archive patterns │ ├── audio-extraction # ffmpeg patterns │ ├── quality-filtering # Accept/reject logic │ ├── metadata-tagging # Tag application │ ├── cover-art-embedding # Artwork embedding │ ├── gap-documentation # GAP-NOTE.md pattern │ ├── integrity-verification # SHA-256 + PREMIS │ └── provenance-tracking # W3C PROV compliance ├── config/ │ └── defaults.yaml # Quality thresholds, export profiles └── docs/ ├── overview.md ├── standards-reference.md └── user-guide.md ``` ### Usage Patterns **Complete collection**: ``` "My favorite band is Radiohead — build me the complete collection" ``` **Targeted extraction**: ``` "I love early Metallica — Master of Puppets era, everything from that period" ``` **Event/tour focus**: ``` "Twenty One Pilots Bandito Tour — every pro-shot performance" ``` ### Export Profiles The framework ships with 5 pre-configured export profiles in `defaults.yaml`: | Profile | Audio Formats | Video | Extras | |---------|--------------|-------|--------| | **Plex** | FLAC, MP3, M4A | MP4, MKV | folder.jpg | | **Jellyfin** | FLAC, Opus, MP3 | MP4, MKV, WebM | folder.jpg, artist.jpg, NFO files | | **MPD** | FLAC, Opus, MP3 | — | folder.jpg | | **Mobile** | Opus, M4A, MP3 | 720p max | 500x500 artwork | | **Archival** | FLAC, WAV | MKV | Checksums + provenance | ## Field-Tested Patterns The framework incorporates learnings from a prototype session cataloging Twenty One Pilots' complete discography (1,109 files, 94GB): ### GAP-NOTE.md Pattern Standardized markdown files placed in directories where content is expected but not yet acquired. They serve as agent instructions, enable parallel gap-filling, and provide completeness measurement by counting remaining gaps. ### opustags over Python mutagen Field testing revealed `opustags` as the superior CLI tool for Opus metadata — faster, more reliable, and better suited for scripted workflows than Python-based alternatives. ### Production-Context Classification Content is classified by production context (studio, live, broadcast) rather than sonic character. This produces more consistent, verifiable categories that align with how media archives are actually organized. ## Standards Compliance - **MusicBrainz** — canonical discography data and Cover Art Archive - **Discogs** — release identification and metadata - **W3C PROV** — provenance tracking for all artifacts - **PREMIS** — preservation metadata for archive integrity - **SHA-256** — fixity verification with self-verifying manifests - **ID3v2 / Vorbis Comments** — audio metadata tagging ## Issues Closed This release closes 10 issues: | Issue | Title | |-------|-------| | #75 | [Epic] Media Curator Framework | | #76 | Discography Analysis Agent | | #77 | Source Discovery System | | #78 | Acquisition Pipeline | | #79 | Quality Assessment Framework | | #80 | Metadata Curation System | | #81 | Completeness Tracking & Gap Analysis | | #82 | Narrative Assembly Tools | | #83 | Multi-Platform Output Preparation | | #253 | Archive Integrity & Fixity Verification | ## Install / Update ```bash npm install -g aiwg@2026.2.7 # Deploy the framework aiwg use media-curator ``` ## Links - [Changelog](https://github.com/jmagly/aiwg/blob/main/CHANGELOG.md) - [Media Curator Docs](https://github.com/jmagly/aiwg/tree/main/agentic/code/frameworks/media-curator/docs) - [User Guide](https://github.com/jmagly/aiwg/blob/main/agentic/code/frameworks/media-curator/docs/user-guide.md)