Vercel/Skills
Properties
- logseq-entity
- Logseq/Entity/Concept
- created-by
- Person/Guillermo Rauch
Vercel Skills Ecosystem
- Vercel's open agent skills platform: a registry of security-audited, installable agent skills discoverable through a web interface and a CLI.
Components
- Vercel/Skills/Web — registry browser at skills.sh; skills browsable by topic (React, Next.js, databases, testing, etc.)
- Vercel/Skills/CLI —
npx skillsinstaller backed by vercel-labs/skills; installs into Claude/Code, Cursor, Codex, GitHub Copilot, Gemini CLI, Cline, and 20+ other tools
Vercel Labs skill repos
Local path source limitation
- My Notes
2026-06-08 Mon inability to use Vercel/Skills to manage skills Declar/at/ive/ly
- tldr
- longer description
npx skills add ./local-pathworks for one-time install butskills-lock.jsonalways stores the absolute path —source-parser.tscallsresolve()on any input before writing the lock file, so relative paths are never stored- Additionally,
update.tsexplicitly skipssourceType: "local"entries duringnpx skills update, so local sources are never updatable via the CLI - Together these make declarative, portable skill installation from a local directory impossible: the lock file encodes a machine-specific absolute path (including the OS username), and the update mechanism ignores it anyway
- This severely limits
vercel/skillsas a declarative install tool for dotfiles repos that manage skills as local source files
- TODO Research whether there are open GitHub issues or PRs on vercel-labs/skills addressing relative path support in
skills-lock.json - Workaround: Until this is resolved, continue using rulesync for skills management in dotfiles — rulesync generates directly into
.claude/skills/from.rulesync/skills/without a lock file abstraction
- My Notes
Ecosystem context
- Competitor to skillshare (cross-agent skills sync) and rulesync (broader scope: rules, commands, subagents, MCP config in addition to skills)
- See AI/Coding/Tool/Report/25/Skills Comparison for comparative analysis; Claude/Code/Skill/Project was the first prominent skills implementation in this ecosystem