agent-factory
Extends the core team with specialists when the goal benefits from domain expertise beyond what the generic agents provide. Runs once after the planner. Never…
This is a reference for the researcher agent. Read the goal and current task, identify which categories apply, then pull sources from those sections only. Ignore irrelevant categories — don't load noise.
> /plugin marketplace add vibhasdutta/loop-engineer > /plugin install loop-engineer@loop-engineer
How it fires
How this agent gets triggered: by you, by Claude, or both.
Context preview
The summary Claude sees to decide when to auto-load this agent.
This is a reference for the researcher agent. Read the goal and current task, identify which categories apply, then pull sources from those sections only. Ignore irrelevant categories — don't load noise.
This is a reference for the researcher agent. Read the goal and current task, identify which categories apply, then pull sources from those sections only. Ignore irrelevant categories — don't load noise.
---
1. Read the goal and task 2. Identify 2–5 relevant categories below 3. Use the listed sources as your primary research targets for that domain 4. For any category not listed here, use general web search to find the best source
---
**General:** Google Search, Brave Search, DuckDuckGo **AI-powered:** Perplexity, Exa Search, Tavily, Phind
---
**Repositories:** GitHub, GitLab, SourceForge **Code Search:** GitHub Code Search, Sourcegraph, grep.app **Discovery:** Awesome Lists (github.com/sindresorhus/awesome), LibHunt, AlternativeTo
---
**Programming:** MDN Web Docs, DevDocs.io, ReadTheDocs **Cloud:** AWS Docs, Google Cloud Docs, Azure Docs, Cloudflare Docs **Frameworks:** The framework's own official docs site (React, Next.js, Vue, Django, etc.) **Languages:** The language's official docs (Python, Rust, Go, Node.js, etc.)
---
**JavaScript:** npm, pnpm registry **Python:** PyPI, Conda **Rust:** crates.io **Go:** pkg.go.dev **Java:** Maven Central **PHP:** Packagist **.NET:** NuGet **Ruby:** RubyGems **Swift:** Swift Package Index
---
**General:** Google Scholar, Semantic Scholar, OpenAlex **CS / AI / ML:** arXiv, Papers With Code, ACM Digital Library, IEEE Xplore **Science:** Nature, ScienceDirect, Springer **Medical:** PubMed, BioRxiv, MedRxiv **Economics / Social:** SSRN, RePec
---
**Models & Weights:** Hugging Face, Ollama Library, OpenRouter, Replicate **Frameworks:** LangChain, LlamaIndex, CrewAI, AutoGen, DSPy, Haystack **Datasets:** Hugging Face Datasets, Kaggle, UCI ML Repository **Benchmarks:** Papers With Code, LMSYS Chatbot Arena, HELM
---
**Discovery:** GitHub (search "mcp-server" or "awesome-mcp"), Smithery, MCP.so **Skills / Plugins:** Platform skill directories, agentskills.io **Integrations:** Official platform docs (Claude, Cursor, Gemini CLI, Antigravity, etc.)
---
**Discovery:** RapidAPI, APIs.guru, Public APIs List (github.com/public-apis) **Testing & Docs:** Postman API Network, Swagger Hub **GraphQL:** GraphQL Hub
---
**Q&A:** Stack Overflow, Stack Exchange **News / Discussion:** Hacker News, Reddit, Lobsters **Project-specific:** GitHub Discussions, GitHub Issues
---
**Blogs:** Dev.to, FreeCodeCamp, DigitalOcean Community, LogRocket Blog, CSS-Tricks **Video:** YouTube (Fireship, Traversy Media, freeCodeCamp) **Courses:** Coursera, Udemy, Khan Academy
---
**Vulnerabilities:** NIST NVD, MITRE CVE, Exploit-DB **Frameworks:** OWASP, MITRE ATT&CK **Threat Intel:** VirusTotal, AlienVault OTX, AbuseIPDB
---
**Patterns:** High Scalability, ByteByteGo, Martin Fowler, InfoQ **Cloud Architecture:** AWS Architecture Center, Azure Architecture Center, Google Cloud Architecture
---
**Company Data:** Crunchbase, LinkedIn, CB Insights **Startups:** Y Combinator Directory, Product Hunt, AngelList (Wellfound)
---
**SEO:** Ahrefs Blog, Semrush, Moz, Backlinko **Content:** HubSpot Blog, Content Marketing Institute
---
**B2B Data:** Apollo, ZoomInfo, Hunter.io, RocketReach **Directories:** Google Maps, LinkedIn, Yelp **B2B Marketplaces:** IndiaMART, Alibaba, TradeIndia
---
**Job Boards:** LinkedIn Jobs, Indeed, Glassdoor, Wellfound **Freelance:** Upwork, Fiverr, Toptal
---
**UI Inspiration:** Dribbble, Mobbin, Awwwards, Behance, Godly **Icons:** Heroicons, Lucide, Font Awesome, Icons8 **Fonts:** Google Fonts, Fontshare
---
**Stock Photos:** Unsplash, Pexels, Pixabay **Illustrations:** unDraw, Storyset **Icons:** Flaticon, Icons8
---
**Platforms:** YouTube, Vimeo **Dev Channels:** Fireship, Traversy Media, freeCodeCamp YouTube
---
**Free:** Project Gutenberg, Internet Archive, Open Library **Technical:** O'Reilly Learning, Google Books
---
**Tech:** TechCrunch, Hacker News, Ars Technica, The Verge **General:** Reuters, AP News, BBC, CNBC
---
**Datasets:** data.gov, data.europa.eu **Legal / Corporate:** SEC EDGAR, Companies House (UK), gov.uk **Patents:** USPTO, WIPO, Google Patents
---
**Markets:** Yahoo Finance, TradingView, Google Finance **Data APIs:** Alpha Vantage, Polygon.io, FRED **Crypto:** CoinMarketCap, CoinGecko
---
**Law:** Cornell Legal Information Institute (LII), Justia, FindLaw **Patents:** WIPO, USPTO, Google Patents, Espacenet
---
**Research:** PubMed, WHO, NIH, CDC **Drugs:** DrugBank, DailyMed
---
**Astronomy:** NASA, ESA **Biology:** NCBI, UniProt **Chemistry:** PubChem **Physics:** CERN, arXiv
---
**Maps:** Google Maps, OpenStreetMap, Bing Maps **Travel & Places:** Wikivoyage, Google Maps
---
**General:** Amazon, eBay **B2B / Wholesale:** Alibaba, AliExpress, IndiaMART **Regional:** Flipkart (India), Walmart
---
**Professional:** LinkedIn **Discussion:** Reddit, X (Twitter), Hacker News **Communities:** Discord, Facebook Groups, Threads
---
**General:** Wikipedia, Britannica, Wikidata **Programming:** ArchWiki, MDN **Domain-specific:** Relevant wiki for the field
---
**Technology Stack:** BuiltWith, Wappalyzer **Archive / History:** Wayback Machine (web.archive.org) **Security:** SS
Loop engineering skill for AI — scaffold a 8-agent team that discovers, implements, verifies, and iterates until your goal is met.
Repo: vibhasdutta/loop-engineer
Extends the core team with specialists when the goal benefits from domain expertise beyond what the generic agents provide. Runs once after the planner. Never…
Catches problems the verifier wouldn't — things that work but aren't done the right way. Uses researcher-defined quality standards to judge output.…
Completes exactly one task and moves the loop forward. Derives execution method from the goal and researcher findings. Output goes to the project directory,…
- Google Search - Bing Search - Brave Search - DuckDuckGo - Kagi - Mojeek - Yahoo Search - Yandex - SearXNG - Startpage