curate-toolkit

Rank installed AI tools by project relevance and write tiered overlay reports.

Updated May 25, 2026
One-click install
npx skills add https://github.com/DATA-AI-XYZ/Tandem --skill curate-toolkit-data-ai-xyz
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: curate-toolkit
Source: https://github.com/DATA-AI-XYZ/Tandem/tree/main/skills/curate-toolkit
Command: npx skills add https://github.com/DATA-AI-XYZ/Tandem --skill curate-toolkit-data-ai-xyz

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When you have many Skills, Agents, Commands, and Plugins installed, it is hard to know which ones actually fit your current project and which are off-stack noise. This Skill audits the installed inventory against your project's type and tech stack, then produces a ranked HIGH / MED / LOW relevance report so you know which tools to reach for and which to deprioritise. ## Core Features & Use Cases - Consumer-first inventory enumeration: Walks user-level installs, the installed-plugins registry, plugin-bundled items, and repo-local overrides in a fixed order so nothing installed is missed. - Judgment-led ranking: Assigns HIGH / MED / LOW tiers with a one-line rationale keyed to project type, stack, and sub-agent mapping, with kit-native items ranked first by default. - Overlay report generation: Writes a dated relevance overlay under _00-Project-Management/97-AI-Reference/ conforming to the ADR-0029 schema, including explicit gap markers for referenced-but-uninstalled items. - Use Case: After installing several new plugins, run the audit to refresh the overlay and see which tools match your Next.js web-app and which are off-stack, then use the HIGH-ranked agents as your preferred executor pool. ## Quick Start Ask the assistant to rank my installed AI tools for this project and write the relevance overlay using /tandem:curate-toolkit.

Frequently Asked Questions about curate-toolkit

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

FAQPage Schema
How do I rank installed Claude Code skills and agents for my project?

Run the curate-toolkit command, which reads PROJECT-CONTEXT.md for your project type and stack, enumerates all installed Skills, Agents, Commands, and Plugins, and assigns each a HIGH, MED, or LOW tier with a one-line rationale written to a dated overlay file.

How does the tool inventory enumeration work?

It walks four steps in order: user-level installs under ~/.claude, the installed-plugins registry, each plugin's bundled skills and commands, then repo-local paths as overrides. A self-test verifies kit-native items appear before any ranking is written.

What happens if a referenced agent or skill is not installed?

Uninstalled items referenced in frontmatter, the sub-agent map, or story files are recorded as explicit gap markers with installed: false, rank LOW, and a GAP-prefixed rationale. Gaps are informational and never abort the ranking pass.

Where are the relevance overlay files written?

Overlays are written to _00-Project-Management/97-AI-Reference/ as curate-toolkit-<YYYYMMDD>.md, with numeric suffixes for same-day reruns. The directory is created if absent, and files follow the ADR-0029 v1 schema or an interim format until it lands.

Can a kit-native tandem skill be ranked below HIGH?

Kit-native items default to HIGH and are listed first, but one can be demoted if PROJECT-CONTEXT.md explicitly excludes the workflow it serves. The justification must be recorded in the item's rationale so future runs see the reasoning.