SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillspi-trending

pi-trending

Agents & models

View repository

Updated 10-15 19:50

About this skill

Pi Trending

Scan the npm registry to discover recently 🔥 Pi Agent packages (extension / skill / theme / prompt template), calculate trend scores, and display a summary to help users track the latest developments in the pi ecosystem.

Output two independent rankings:

  • Mainstream ranking - Sorted descending by monthly downloads over the past 30 days (monthly), reflecting the packages currently most used in the ecosystem
  • Emerging ranking - Sorted descending by growth score (growth × ln(weekly+1)), reflecting packages that are rapidly rising

Quick start

Run one command to get both the mainstream ranking and the emerging ranking as two Markdown tables (LLM-friendly). With no arguments, each shows Top 30.

uv run --script scripts/pi_trending.py

AI display stepWhen the AI gets the table output by the script, it should automatically translate the English description in the "One-sentence introduction" columninto Chinese and condense it to within 20 characters, then display it to the user. Descriptions that fail to translate keep the original text.

Example script output (description is the original English; before display, translate the "One-sentence introduction" column according to the rules above):

# 🔥 Pi Agent 最新热门包 (2026-06-17)

## 主流榜
| # | 包名 | 作者 | 月下载量 | 一句话介绍 |
|---|------|------|---------|------------|
| 1 | `@pi/core` | user | 12,345 | A core framework for building pi extensions |

> Top 30 · 按月下载量排序

## 新锐榜
| # | 包名 | 作者 | 趋势分 | 一句话介绍 |
|---|------|------|--------|------------|
| 1 | `pi-context-map` | dev | 9 | Professional context profiler for Pi |

Ranking notes

Data source: npm registry (npm packages containing the pi-package keyword). Search strategy:sort=popularity collect 500 candidate packages from the npm search API sorted by downloads. Core types:extension · skill · theme · prompt(automatically identified via npm keywords).

Mainstream ranking

Sorted descending by downloads.monthly(monthly downloads over the past 30 days returned by the search API), with zero extra API cost. Reflects thepackages currently most used, suitable for answering "what is everyone using".

Emerging ranking

Sorted descending by growth score, reflectinggrowth speedrather than absolute popularity. The algorithm uses only the weekly/monthly two fields returned by the search API:

recency = weekly / monthly                       # 本周在月度总量中的占比
growth  = (recency - 7/30) / (1 - 7/30)          # 归一化到 0~1，7/30 为均匀分布基线
score   = growth × ln(weekly + 1)                # 增速 × 信誉权重
  • recency ≤ 7/30packages whose share this week is not above the uniform-distribution baseline score 0 and do not enter the ranking
  • Candidate threshold:weekly ≥ 100, below which they do not enter the emerging ranking candidate pool (to avoid low-base noise)

Score magnitude reference (at the pi ecosystem's weekly scale,growth ≤ 1 × ln(weekly+1) upper limit is about 11):

Trend score Meaning
≥8 🚀 This week's share far exceeds the baseline, explosive growth
≥3 🔥 Clearly above the baseline, rising rapidly
>0 📈 Slightly above the baseline, moderate growth
0 Not ranked (recency ≤ 7/30 or weekly < 100）

⚠️ Notes：

  • New packages weekly have a low hourly trend score (ln(weekly+1) items are small), which does not mean there is no growth
  • Mature and stable large packages (such as core frameworks) rank high in the mainstream ranking, but their trend score in the emerging ranking may not be high
  • Trend scorereflects growth speed; it is recommended to evaluate it together with the weekly_downloads in the JSON output (--json）
  • The same package can appear in both rankings at the same time (they do not conflict)

CLI reference

uv run --script scripts/pi_trending.py [选项]
Option Effect Default
--type <extension|skill|theme|prompt> Filter the candidate pool by type; both rankings show only that type all
--max N Set the number of entries for both the mainstream and emerging rankings 30
--mainstream-max N Set only the number of mainstream ranking entries (overrides --max） same as --max
--rising-max N Only set the number of rising-star entries (overrides --max); set to 0 to read only the mainstream list Same as --max
--json JSON output (including list_type field to distinguish mainstream/rising), can be piped to jq off
--verbose Output API request progress, pagination, candidate counts, etc. to stderr off

JSON pipeline example:

# 只读主流榜包名
uv run --script scripts/pi_trending.py --json | jq '.[] | select(.list_type == "mainstream") | .name'

# 只读新锐榜包名
uv run --script scripts/pi_trending.py --json | jq '.[] | select(.list_type == "rising") | .name'

Next steps

The package names in the trend table are npm packages, and can be used directly with npm/gh to trace details and changes.

View package details：npm view <包名> readme(including @scope/ prefix).

Install(by package type):

  • pi-extension -> bunx pi extension install <name>
  • pi-skill -> bunx skills add <name> or bunx skills add <gh-repo>
  • pi-theme -> bunx pi theme install <name> or refer to the theme installation guide
  • prompt-template -> bunx pi prompt install <name> or write to the prompt-templates directory

View recent changes(packages rising quickly in the trend table are often worth following up on): use npm view <包名> time --json to see release cadence,npm view <包名> repository to find the repository, then use gh release view / gh api repos/<owner>/<repo>/commits to view release notes and commits. Analysis points:

  • Version release timing density -> development activity signal
  • Version number jumps (e.g. 1.x -> 2.0) -> major refactor or production hardening
  • Follow up on a specific version number -> npm view <包名> readme view the changelog section of the README

Script location

scripts/pi_trending.py - uv-based PEP 723 single-file script, automatically manages dependencies.