skills.rest
MeridianVega
MeridianVegaOFFICIAL
Explore all skills by this author →
WebsiteGitHub

librarian-taste

Create and update a versioned taste.md profile with append-only feedback.

Updated Aug 23, 2026
Search Tags:#interview#session-management#survey#content-discovery#librarian#taste-profile#taste.md
One-click install
npx skills add https://github.com/MeridianVega/claude-marketplace --skill librarian-taste
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: librarian-taste
Source: https://github.com/MeridianVega/claude-marketplace/tree/main/plugins/ersatztv-programmer/skills/librarian-taste
Command: npx skills add https://github.com/MeridianVega/claude-marketplace --skill librarian-taste

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill creates and maintains a persistent content-taste profile (taste.md) so the system can choose content aligned with your preferences and past feedback. It governs how you sample library signals and how your preferences persist across sessions.

Core Features & Use Cases

  • Builds the taste.md schema and versioned profile, implementing an "Always include", "Never include", "Tilt toward" structure based on your inputs, library signals, and feedback.
  • Supports two paths: library survey + targeted questions when Jellyfin/ library data exists, and a pure interview path when the library is absent or small.
  • Append-only feedback tracking (Recent confirmed loves/misses) to evolve the profile over time without overwriting user-stated preferences.

Quick Start

Create your first taste.md by initiating a guided session and confirming the draft.

Frequently Asked Questions about librarian-taste

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

FAQPage Schema
How do I build a persistent content-taste profile for session management?▼

You start a guided session to create a taste.md file, which builds a versioned content-taste profile using an 'Always include', 'Never include', and 'Tilt toward' structure based on your interview inputs and library signals.

What is the best way to track content feedback without overwriting past preferences?▼

The best way to track content feedback is an append-only system that logs confirmed loves and misses to your taste profile, evolving it across sessions without overwriting user-stated preferences.

Does the taste profile interview work without Jellyfin library data?▼

Yes, the taste profile interview works without Jellyfin library data by falling back to a pure interview path, whereas a library survey with targeted questions is used when Jellyfin or library data exists.

Can I review and confirm my taste profile before it saves?▼

Yes, you can review your taste profile before it saves, as the system requires explicit user confirmation of the draft taste.md file before persisting any updates from the survey or interview session.

How does versioning work in a taste.md content discovery profile?▼

Versioning in a taste.md content discovery profile works by applying a strict schema that separates past loves and misses, using append-only updates to evolve preferences over time while maintaining historical session data.

Related Skills

Explore More Content & Communication Skills→
lemeblemeb

interview

Conduct structured feature interviews and output a formal specification document.

Community
Advanced
cxnaivecxnaive

here.now

Publish static websites and manage private cloud storage drives via curl, file, and jq.

Community
Intermediate
magnus919magnus919
40

jellyfin-cli

Query Jellyfin REST APIs to retrieve library metadata and search results.

Community
Intermediate
0xHoneyJar0xHoneyJar
2

sigil

Record taste preferences to guide design physics in Rune.

Official
Basic