aboylabs/ai-agent-skills / skills/research/trending-repos-discovery
trending-repos-discovery
VerifiedDiscover and analyze trending GitHub repositories from TrendShift and other sources β evaluate usefulness, extract learnings, and identify valuable skills/tools
Trending Repos Discovery
Discover and analyze trending GitHub repositories from TrendShift and other sources. Evaluate usefulness, extract learnings, and identify valuable skills/tools for adoption.
When to Use
- User asks "apa yang trending?" or "what's trending?"
- User wants to discover new tools/libraries
- User asks about specific trending repo
- Looking for inspiration or best practices
- Evaluating whether to adopt new tools
Sources
1. TrendShift (Primary)
What it tracks:
- GitHub's most-discussed open-source projects
- Daily engagement rankings
- Real-time mentions from X/Twitter
- Discussions from Reddit and Hacker News
How to fetch: See references/trendshift-fetch-methods.md for TinyFish and GitHub API examples
2. GitHub Trending (Alternative)
URL: https://github.com/trending
Languages:
- All languages: https://github.com/trending
- Python: https://github.com/trending/python
- TypeScript: https://github.com/trending/typescript
- Rust: https://github.com/trending/rust
3. GitHub API (For Details)
See references/trendshift-fetch-methods.md for API usage examples.
Rate limit: 60 requests/hour (unauthenticated), 5000/hour (authenticated)
Analysis Workflow
Phase 1: Fetch Trending List
- Fetch from TrendShift (primary source)
- Extract repo names (owner/repo format)
- Get first 10-20 repos (manageable batch)
Phase 2: Get Repo Details
For each repo, fetch from GitHub API:
- Stars, forks, watchers
- Description
- Language
- Topics
- Created/updated dates
Batch strategy:
- Fetch 10 repos at a time
- If rate limited, wait or use cached data
Phase 3: Categorize by Usefulness
For each repo, assess:
A. Relevance Score
| Factor | Weight | Criteria |
|---|---|---|
| Stars | 30% | > 10K = high, 1K-10K = medium, < 1K = low |
| Description match | 30% | Keywords match user's interests |
| Language | 20% | User's tech stack (Python, TypeScript, Rust) |
| Recency | 10% | Updated in last 30 days |
| Topics | 10% | Relevant tags (ai, agent, skills, tools) |
B. Usefulness Categories
β Highly Useful:
- Directly applicable to user's projects
- Solves current problems
- Fills gaps in toolset
- High stars (> 10K) + active development
π‘ Potentially Useful:
- Interesting but not immediate need
- Requires evaluation/testing
- Medium stars (1K-10K)
- Niche use case
β Not Useful:
- Wrong tech stack (e.g., Mac-only for Linux user)
- Too niche (e.g., biology tools for software dev)
- Duplicate of existing tools
- Low stars (< 1K) + no clear value
Phase 4: Extract Learnings
For useful repos, identify:
1. Skills/Patterns
- What skills does it provide?
- What patterns does it use?
- What best practices does it demonstrate?
Example (obra/superpowers):
- Skills framework architecture
- Agentic workflow patterns
- Production-ready methodology
2. Tools/Libraries
- What tools does it introduce?
- What dependencies does it use?
- What integrations does it support?
Example (addyosmani/agent-skills):
- Production-grade engineering patterns
- Web performance optimization
- Chrome DevTools expertise
3. Concepts/Ideas
- What problems does it solve?
- What approaches does it take?
- What innovations does it introduce?
Example (andrej-karpathy-skills):
- LLM coding pitfalls
- AI code quality patterns
- Production AI best practices
Phase 5: Present Analysis
Format:
## π₯ TOP TRENDING REPOS
### 1. owner/repo β X,XXX stars π
**Description:** [One-line description]
**Berguna?** β
SANGAT / π‘ MUNGKIN / β TIDAK
**Why:**
- [Reason 1]
- [Reason 2]
- [Reason 3]
**Use Case:** [How user can use it]
**Action:** [What to do next]
---
## π― RECOMMENDED ACTIONS
### Priority 1: MUST CHECK π₯
1. [Repo 1] - [Why]
2. [Repo 2] - [Why]
### Priority 2: WORTH CHECKING β
3. [Repo 3] - [Why]
### Priority 3: MAYBE β οΈ
5. [Repo 5] - [Why]
## β SKIP THESE
- [Repo X] - [Why not useful]
User Preference Patterns
For "Analisa trending repos"
When user asks to analyze trending repos:
DO:
- β Fetch from TrendShift (most comprehensive)
- β Get details for top 10-20 repos
- β Categorize by usefulness (β π‘ β)
- β Explain WHY each is useful/not useful
- β Provide actionable recommendations
- β Group by priority (must check, worth checking, maybe, skip)
DON'T:
- β Just list repo names (no value)
- β Skip usefulness assessment
- β Recommend everything (no filtering)
- β Ignore user's tech stack/constraints
For "Ada yang berguna gak?"
When user asks if trending repos are useful:
Response pattern:
- Quick verdict: "ADA YANG BERGUNA? β BANYAK!" or "β TIDAK ADA"
- Top picks: List 3-5 most useful repos
- Why useful: Explain specific benefits
- Action: What to do next (clone, study, adapt)
Example:
β
ADA YANG MENARIK β review dulu sebelum klaim apapun
1. obra/superpowers β Skills framework
2. andrej-karpathy-skills β coding guidelines
3. mattpocock/skills β TypeScript / engineering workflows
4. addyosmani/agent-skills β frontend & performance practices
Mention star counts only when reporting raw findings; do not aggregate
them into "X K combined stars" or claim authors as contributors to a
downstream repo. Authors of upstream public skill repos have not endorsed
or contributed to any aggregator unless they explicitly say so.
MAU GUE CLONE & ANALISA SEKARANG? π
Evaluation Criteria
For Skills Repos
Look for:
- β SKILL.md or .claude directory (skills format)
- β High stars (> 10K = trusted)
- β Industry expert author (Karpathy, Pocock, Osmani)
- β Production-tested (not toy examples)
- β Applicable to user's work (coding, AI, web dev)
Red flags:
- β Low stars (< 1K) + unknown author
- β Toy examples only
- β Outdated (not updated in 6+ months)
- β Wrong tech stack (e.g., Java for Python user)
For Tools/Libraries
Look for:
- β Solves real problem
- β Active development (commits in last 30 days)
- β Good documentation (README, examples)
- β Compatible with user's stack
- β Resource cost acceptable (< 500 MB disk, < 100 MB RAM for VPS users)
Red flags:
- β Heavy resource usage (> 500 MB disk, > 100 MB RAM)
- β Platform-specific (Mac-only for Linux user)
- β Duplicate of existing tools
- β Poor documentation
For Frameworks
Look for:
- β Clear architecture
- β Best practices demonstrated
- β Production-ready
- β Extensible/adaptable
- β Good examples
Red flags:
- β Over-engineered
- β Too opinionated
- β Hard to adapt
- β No examples
Common Patterns
Skills Repos (High Value)
Pattern:
- Name:
*-skills,skills,agent-skills - Format: SKILL.md, .claude directory, or markdown files
- Author: Industry experts (Karpathy, Pocock, Osmani)
- Stars: > 10K
- Content: Production-tested patterns, best practices
Examples:
- obra/superpowers (186K stars) - Skills framework
- forrestchang/andrej-karpathy-skills (125K stars) - Coding quality
- mattpocock/skills (73K stars) - TypeScript mastery
- addyosmani/agent-skills (39K stars) - Performance
Action:
- Clone repo
- Study SKILL.md files
- Identify useful patterns
- Adapt to Hermes skills
- Test in projects
Tool Repos (Medium Value)
Pattern:
- Name: Tool/library name
- Format: Python package, npm package, CLI tool
- Stars: 1K-10K
- Content: Specific functionality
Examples:
- antirez/ds4 (7.5K stars) - DeepSeek inference (Mac-only)
- huangserva/3DCellForge (1.5K stars) - 3D cell generation (niche)
Action:
- Check resource cost (disk, RAM)
- Check compatibility (OS, dependencies)
- Evaluate usefulness (solves problem?)
- If useful + lightweight β install
- If heavy β find alternative
Framework Repos (High Value if Applicable)
Pattern:
- Name: Framework name
- Format: Full project structure
- Stars: > 10K
- Content: Architecture, patterns, examples
Examples:
- github/spec-kit (96K stars) - Spec-Driven Development
Action:
- Study architecture
- Understand patterns
- Evaluate applicability
- Adapt concepts (not full framework)
- Document learnings
Pitfalls
1. Recommending Everything
Problem: User asks "ada yang berguna?", you list all 20 repos.
Solution:
- Filter by usefulness (β π‘ β)
- Recommend top 3-5 only
- Group by priority
- Explain why each is useful
2. Ignoring Resource Constraints
Problem: Recommend heavy tool (> 500 MB) for VPS user.
Solution:
- ALWAYS check resource cost first
- If heavy β find lightweight alternative
- If no alternative β skip
- See
vps-cleanupskill for thresholds
3. Not Explaining Why
Problem: List repos without explaining usefulness.
Solution:
- Always explain WHY useful
- Specific benefits (not generic "good tool")
- How user can use it
- What problem it solves
4. Recommending Wrong Tech Stack
Problem: Recommend Mac-only tool for Linux user.
Solution:
- Check compatibility (OS, dependencies)
- Check user's tech stack (Python, TypeScript, Rust)
- Skip incompatible tools
- Mention incompatibility in analysis
5. Not Providing Next Steps
Problem: User says "mantap" but doesn't know what to do.
Solution:
- Always end with action items
- "MAU GUE CLONE & ANALISA SEKARANG?" (Want me to clone & analyze now?)
- Specific commands (git clone, study, adapt)
- Clear next steps
Examples
See references/top-4-skills-repos-case-study.md for detailed example from session 2026-05-12.
References
- TrendShift: https://trendshift.io/
- GitHub Trending: https://github.com/trending
- GitHub API: https://docs.github.com/en/rest
- Fetch methods:
references/trendshift-fetch-methods.md - Case study:
references/top-4-skills-repos-case-study.md
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why β repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" β or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself β your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |