mcpField
B
best-skills

best-skills

github.com/linklyaiAgent workflows67247 installsUpdated 1d ago
Repository

About best-skills

best-skills aggregates rankings and per-platform counts for Agent skills across multiple communities, with daily updates. It helps users discover and compare listed skills, but does not install them or search the entire skills universe.

Key features

  1. 01Daily updates to nine Top 100 lists
  2. 02Per-platform raw counts
  3. 03Trend and rising-skill rankings
  4. 04Skill comparison by key
  5. 05Historical ranking snapshots

Use cases

  1. 01Filter for browser automation or PDF skills
  2. 02Check recent trends and rising skills
  3. 03Compare two skills and their platform counts
  4. 04Check whether a skill appears in official or active lists

MCP support

The input does not specify a relationship to MCP.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/linklyai/best-skills --skill best-skills

After installing, type/best-skillsin Claude Code to confirm it loaded.

Details

From the project README

Best Agent Skills

Nine Top-100 lists, refreshed daily, with the raw per-platform numbers kept next to every score. Full guide and column dictionary: https://raw.githubusercontent.com/LinklyAI/best-skills/main/llms.txt

When to use

  • "Which skill should I install for browser automation / PDF / …" → best-100.csv, filter by description and skill
  • "What is trending / new this week" → trending-7d.csv, rising-stars.csv
  • "Is <skill> popular, maintained, official?" → best-100.csv (score and dimensions), official-100.csv, most-active.csv
  • "Compare A and B" → look both up by skill_key, cite raw counts per platform

Not for: installing skills (hand the user the install command), full-text search of every skill in existence (only the Top 100 per list is published; the raw snapshots under data/latest/raw/ go deeper), judging quality beyond the published counts.

Workflow

  1. Fetch https://raw.githubusercontent.com/LinklyAI/best-skills/main/llms.txt first. It is the current column dictionary and the rules; do not rely on a cached copy.
  2. Pick the list, fetch https://raw.githubusercontent.com/LinklyAI/best-skills/main/data/latest/rankings/<list>.csv (4–90 KB, ≤100 rows). Parse by header name; an empty cell means "not measured", not zero.
  3. Answer with rank, the raw numbers, install and url. Show vendor next to skill: same-name skills from different vendors are different skills.
  4. For history, replace latest with YYYY-MM-DD (daily since 2026-08-16).

Lists

best-100 (overall score), top-installs, trending-7d, social-buzz, most-active, official-100, official-vendors, top-repos, rising-stars.

Rules

  • Never add install counts across platforms; present them side by side.
  • Absent from a list ≠ zero; lists are truncated at 100.
  • Attribute republished data: "Data from Best Skills (https://github.com/LinklyAI/best-skills) — CC BY 4.0".

Comments

Markdown supported