开源项目评分器

Search, evaluate, and curate GitHub open source projects with personalized scoring and deduplication.

Overview

This skill extends basic GitHub search with a structured evaluation pipeline: user preferences define scoring criteria, each project is scored across multiple dimensions, and only projects meeting the threshold are recorded — with deduplication to avoid revisiting previously rated projects.

Core Capabilities

1. Preference Management

Read references/preferences.md at the start of each session. This file contains:

• Project interest areas (or instructions to use per-query topics)
• Scoring dimensions (Innovation, Popularity, Code Quality, Maintenance)
• Dimension weights and passing threshold
• Customization rules

When the user requests changes to preferences (e.g., "add a dimension", "change the threshold", "adjust weights"), update references/preferences.md immediately and confirm.

2. Project Search

Use scripts/search_github.js to query the GitHub API:

# Flag-based interface (recommended)
scripts/search_github.js -q "<query>" [-n max_results] [-l language] [-s sort] [--date-from YYYY-MM-DD]

# Examples:
scripts/search_github.js -q "LLM inference optimization" -n 10 -l python
scripts/search_github.js -q "web framework" -n 20 -l javascript -s stars
scripts/search_github.js -q "docker kubernetes" -n 15

Flag arguments:

| Flag | Short | Default | Description | |------|-------|---------|-------------| | --query | -q | (required) | Search keywords | | --max-results | -n | 10 | Number of results to return | | --language | -l | none | Programming language filter, e.g. python, javascript, go, rust | | --sort | -s | stars | Sort by: stars, updated, created | | --date-from | — | none | Filter repos pushed/updated after this date |

Parse the returned JSON. Each repository contains full_name, description, stargazers_count, fork_count, language, updated_at, topics, and html_url.

3. Project Scoring

For each project found, perform the following evaluation:

1. Read the project info (name, description, topics)
2. Check deduplication: Search references/rated_projects.md for the project's GitHub ID (full_name). If found, skip and note as duplicate.
3. Score each dimension (1-10) per the rubric in references/preferences.md:
   • Innovation (创新性) — 是否提出新思路或独特解决方案
   • Popularity (热度) — stars、fork数量及增长趋势
   • Practicality (实用性) — 文档完善度、使用场景、能否直接采用
   • Maintenance (活跃度) — 最近更新时间、Issue处理速度、社区活跃度
4. Calculate weighted score using the weights from preferences
5. Determine status:
   • Pass: Weighted Score >= threshold (default 6.5)
   • Borderline: Within 1.0 point below threshold
   • Fail: Below borderline range

4. Record Keeping

For projects that pass the scoring threshold, append an entry to references/rated_projects.md using this format:

### [SCORE] GitHub_ID - PROJECT_NAME
- **Author**: owner/repo
- **Language**: Primary language
- **Stars**: NNNNN
- **Rated**: YYYY-MM-DD
- **Link**: https://github.com/owner/repo
- **Scores**: Innovation=N Popularity=N Practicality=N Maintenance=N
- **Weighted**: N.NN
- **Keywords**: topic1, topic2
- **Summary**: One-sentence summary.

Deduplication: Before scoring any project, check references/rated_projects.md for an existing entry with the same GitHub ID (owner/repo). If found, skip the project and inform the user it was previously rated.

Workflow

Standard Search & Rate Workflow

1. Read references/preferences.md to load current preferences and scoring criteria
2. Read references/rated_projects.md to load previously rated project IDs
3. Accept user's search topic/keywords (or use default interests from preferences)
4. Run scripts/search_github.js -q "<query>" -n <count> [-l language]
5. Parse JSON results, extract project entries
6. For each project: a. Extract GitHub ID from full_name b. Check deduplication against references/rated_projects.md c. If duplicate, skip and note d. If new, score across all dimensions e. Calculate weighted score f. Determine pass/borderline/fail status
7. Present results to the user in a structured table:
   • Project name, language, stars, weighted score, dimension scores, status
   • Highlight passed projects, flag borderline ones
8. For passed projects, append entries to references/rated_projects.md
9. Optionally provide a brief analysis of the batch results

Preference Update Workflow

When the user requests preference changes:

1. Read current references/preferences.md
2. Apply the requested changes
3. Write updated references/preferences.md
4. Confirm the changes to the user

History Review Workflow

When the user wants to review previously rated projects:

1. Read references/rated_projects.md
2. Present the entries, optionally filtered by keyword, language, or score range

Output Format

Present search results in a clear table (must use Chinese table headers):

| # | 项目 | 语言 | Stars | 综合分 | 创新 | 热度 | 实用 | 活跃 | 状态 | |---|------|------|-------|--------|------|------|------|------|------| | 1 | owner/repo | Python | 12.5k | 7.8 | 8 | 9 | 7 | 7 | PASS | | 2 | owner/repo2 | Go | 3.2k | 6.2 | 7 | 6 | 6 | 5 | BORDERLINE |

说明：创新=Innovation, 热度=Popularity, 实用=Practicality, 活跃=Maintenance

For each passed project, also provide a 2-3 sentence summary explaining why it passed and what makes it noteworthy.

Examples

• "帮我搜一下最近热门的 LLM 推理优化开源项目，按我的标准评分"
• "Search GitHub for Docker Kubernetes tools, score and filter the results"
• "查看我之前评过的项目"
• "把评分阈值调到 7.0"
• "给热度维度加大权重到 0.35"
• "今天有什么值得关注的 Python 开源项目吗？"

Resources

• scripts/search_github.js: GitHub API search script with language and sort support
• references/preferences.md: User preferences, scoring dimensions, weights, and threshold
• references/rated_projects.md: Record of previously rated projects for deduplication