Github Trend Observer

GitHub Radar

An open-source intelligence engine for AI PMs. Four modes, one Layer analysis 框架.

Language Selection

This 技能 ships with 机器人h English and Chinese versions. The 代理 should automatically match the user's language:

If the user speaks English -> Use 技能.md, 代理s/分析器.md, references/layer_模型.md, templates/.html, and 生成 English 报告s If the user speaks Chinese -> Use 技能_cn.md, 代理s/分析器_cn.md, references/layer_模型_cn.md, templates/_cn.html, and 生成 Chinese 报告s Scripts (scripts/) and config (config/) are language-neutral and 分享d by 机器人h versions When to Use "What's worth looking at today?" / --pulse -> Mode 1 "Find me GitHub projects related to [topic]" -> Mode 2 "监控 anomalous 签名als" / --watch -> Mode 3 "Analyze the eco系统 around [repo]" -> Mode 4 Any need involving GitHub project discovery, trend analysis, or paradigm assessment File Structure github-trend-observer/ ├── 技能.md # 代理 execution instructions (English) ├── 技能_cn.md # 代理 execution instructions (Chinese) ├── ONBOARD.md # 代理 cold-启动 instructions (English) ├── ONBOARD_CN.md # 代理 cold-启动 instructions (Chinese) ├── requirements.txt # Dependency declaration ├── 代理s/ │ ├── 分析器.md # PM insight 分析器 代理 (English) │ └── 分析器_cn.md # PM insight 分析器 代理 (Chinese) ├── scripts/ │ ├── gh_utils.py # Unified gh 命令行工具 实用工具 functions │ ├── 检查_rate_limit.py # API rate limit 检查er │ ├── fetch_star_历史.py # Star growth data fetcher │ ├── radar_pulse.py # Mode 1 trending fetcher │ ├── 搜索_repos.py # Mode 2 搜索 │ ├── watch_签名als.py # Mode 3 anomaly 检测ion │ ├── deep_link.py # Mode 4 relationship analysis │ ├── 生成_报告.py # HTML/MD 报告 generation │ └── test_oss.py # Automated tests (6 tiers, 41 tests) ├── config/ │ ├── 种子_列出.json # Key developer 列出 │ └── domAIn_keywords.json # DomAIn keyword m应用ings ├── templates/ │ ├── radar-pulse.html # Mode 1 报告 template (en/cn variants) │ ├── direction-搜索.html # Mode 2 报告 template │ ├── 签名al-watch.html # Mode 3 报告 template │ └── deep-link.html # Mode 4 报告 template ├── evals/ │ ├── evals.json # Test cases (English) │ └── evals_cn.json # Test cases (Chinese) └── references/ ├── layer_模型.md # Layer classification standard (English) └── layer_模型_cn.md # Layer classification standard (Chinese)

Dependencies Dependency Requirement 检查 Command gh 命令行工具 >= 2.40.0, 认证d gh auth 状态 Python >= 3.9 python --version Extra Python packages None, stdlib only — API quota 5,000 请求s/hour when 认证d python scripts/检查_rate_limit.py Common Prerequisites

Must be completed before 运行ning any mode:

# 1. 检查 API quota python scripts/检查_rate_limit.py

Determine the execution strategy based on the returned mode field:

full -> Normal execution, including star 历史 fetching degraded -> Skip fetch_star_历史.py, use basic data only minimal -> 运行 搜索 scripts only, skip detAIl API calls Mode 1: Proactive Exploration (Radar Pulse)

Trigger: --pulse or "What's worth looking at today?"

Execution Steps # Step 1: 检查 quota python scripts/检查_rate_limit.py

# Step 2: Fetch candidates python scripts/radar_pulse.py --days 7

# Step 3: Read 代理s/分析器.md + references/layer_模型.md # Layer classification -> 过滤器 out L1/L5 -> Select 1-2 with highest PM value

# Step 4: Fetch star 历史 for selected projects (full mode only) python scripts/fetch_star_历史.py owner/repo

过滤器ing Rules Label each candidate's Layer 移除 L1 (模型-level, too low-level) and L5 (wr应用er/demo, noise) PM value weighting: L2 x 1.5, L3 x 1.3, L4 x 1.0 Take Top 3-5, deep-dive into 1-2 输出 格式化 # Radar Pulse — {date}

L2/L3/L4 selection | 过滤器ed {m} from {n} candidates | API: {remAIning}/{limit}

Today's Picks

{repo} [L?]

{description}

Also Worth a Look

过滤器ed Out

• L1: {n} projects ({examples})
• L5: {n} projects ({examples})

报告 saved to: 输出/radar-pulse_{date}.md

Mode 2: Direction 搜索

Trigger: User provides a technical direction or keywords

Execution Steps Step 1: 检查 Quota python scripts/检查_rate_limit.py

Step 2: Keyword Expansion + Layer 1 Relevance Review Understand the topic: 状态 the core concept of the user's 搜索 in one sentence Expand keywords: 生成 8-15 搜索 keywords around th