docs / cat.md
docs / cat.md
Plugins designed to connect ready-made MCP servers or third-party proxy gateways for aggregating multiple AI providers without reinventing the wheel.
@dangahagan/weather-mcp, @modelcontextprotocol/server-filesystem), hoonowng/qemu-mcp (QEMU VM control with 182 built-in MCP tools), cyb3rjak3/cloudflare-ai-gateway (Cloudflare AI proxy provider), count-greebo/perplexity-search.Tools for reading, editing, searching through local files, and automatically converting document formats (including PDFs and Excel sheets).
lmbimmerboye/file-agent, fuutott/save-excel (saving tables directly to files), zexigh/read-pdf (OCR-scanning PDF to text), bakit/ai-to-pdf (PDF conversion via browser under Puppeteer), trebuchet/small-file-agent (search, read, and replace text in files).Plugins that help the AI retain details across different sessions, manage memory seeds, handle associations, and inject context dynamically.
lanicorec/alexiel-memory (built-in SQLite storage), anhuvdinh/persisting-memories (preserving reusable memory seeds), chokito/persistent-memory, student489/kv-store (Key-Value store inspired by persistent memory), tupik/memory (simple JSON slot-based memory).Plugins for scanning web pages, bypassing search engine limits, and injecting up-to-date internet content straight into the model's active context.
danielsig/duckduckgo (deep web search with CAPTCHA handling and post requests), danielsig/visit-website (pulling text, links, and images by URL), rzk/searxng-search (local private search engine wrapper), artemyershov/fetcher (web scraper), nightly-raven/reddit-search.Plugins that automatically compress, prune, or summarize dialogue history when the conversation edges close to the model's token limits.
rootmonster/context-compressor (auto-compress triggered at 3,000 tokens), alexandreaxell/context-compactor (summarizing the earliest parts of the chat), jorbraken/brain (context-cleanup utilities).Giving the AI agent direct access to the host machine's command line, sandboxed Python code runners, or isolated operating systems.
khtsly/computer (AI-controlled virtual machine via WSL2/Docker with Ubuntu 24), khtsly/terminal (shell execution for running bash scripts), cseliot/shell-command-runner, soumyajit7038/python-tools (dedicated environment to execute Python code).Vector search integrations designed to parse user knowledge bases, personal markdown vault setups, and large digital document libraries.
lmstudio/rag-v1 and dirty-data/rag-v2, mindstudio/big-rag (optimized for Obsidian Vaults), phuocnguyen90/omnimind (RAG for Obsidian and Zotero PDFs), johndoe777/context-aware-rag-rus.Local servers that emulate the behavior of commercial APIs (like OpenAI), allowing seamless integration of third-party software (e.g., VS Code extensions) with local LLMs.
openai-compat-endpoint fork family (OpenAI API compatibility shim for external clients), Qwen Code integration with VS Code code companion plugins using a local model setup.Quality-of-life add-ons that automate minor tasks such as exporting chats, injecting dynamic timestamps, processing basic calculations, or handling reminders.
thefuchsen/time-injector (dynamic watch/timestamp for AI context), promptpirate/chat-auto-exporter (exporting dialogue chunks every 10 steps), dimasbotyara/calculator, desumor/reminder-llm.Tools built for response auditing, LLM penetration testing, scrubbing Personally Identifiable Information (PII), and checking texts for AI generation.
zexigh/anonymize (scrubbing PII before sending prompts), altra/humanize (scoring the natural flow of AI-generated text), glyphsoftware/sentinel-r2.1-gguf (proprietary LLM toolset for pentesting and censorship checking).Plugins designed to connect ready-made MCP servers or third-party proxy gateways for aggregating multiple AI providers without reinventing the wheel.
@dangahagan/weather-mcp, @modelcontextprotocol/server-filesystem), hoonowng/qemu-mcp (QEMU VM control with 182 built-in MCP tools), cyb3rjak3/cloudflare-ai-gateway (Cloudflare AI proxy provider), count-greebo/perplexity-search.Tools for reading, editing, searching through local files, and automatically converting document formats (including PDFs and Excel sheets).
lmbimmerboye/file-agent, fuutott/save-excel (saving tables directly to files), zexigh/read-pdf (OCR-scanning PDF to text), bakit/ai-to-pdf (PDF conversion via browser under Puppeteer), trebuchet/small-file-agent (search, read, and replace text in files).Plugins that help the AI retain details across different sessions, manage memory seeds, handle associations, and inject context dynamically.
lanicorec/alexiel-memory (built-in SQLite storage), anhuvdinh/persisting-memories (preserving reusable memory seeds), chokito/persistent-memory, student489/kv-store (Key-Value store inspired by persistent memory), tupik/memory (simple JSON slot-based memory).Plugins for scanning web pages, bypassing search engine limits, and injecting up-to-date internet content straight into the model's active context.
danielsig/duckduckgo (deep web search with CAPTCHA handling and post requests), danielsig/visit-website (pulling text, links, and images by URL), rzk/searxng-search (local private search engine wrapper), artemyershov/fetcher (web scraper), nightly-raven/reddit-search.Plugins that automatically compress, prune, or summarize dialogue history when the conversation edges close to the model's token limits.
rootmonster/context-compressor (auto-compress triggered at 3,000 tokens), alexandreaxell/context-compactor (summarizing the earliest parts of the chat), jorbraken/brain (context-cleanup utilities).Giving the AI agent direct access to the host machine's command line, sandboxed Python code runners, or isolated operating systems.
khtsly/computer (AI-controlled virtual machine via WSL2/Docker with Ubuntu 24), khtsly/terminal (shell execution for running bash scripts), cseliot/shell-command-runner, soumyajit7038/python-tools (dedicated environment to execute Python code).Vector search integrations designed to parse user knowledge bases, personal markdown vault setups, and large digital document libraries.
lmstudio/rag-v1 and dirty-data/rag-v2, mindstudio/big-rag (optimized for Obsidian Vaults), phuocnguyen90/omnimind (RAG for Obsidian and Zotero PDFs), johndoe777/context-aware-rag-rus.Local servers that emulate the behavior of commercial APIs (like OpenAI), allowing seamless integration of third-party software (e.g., VS Code extensions) with local LLMs.
openai-compat-endpoint fork family (OpenAI API compatibility shim for external clients), Qwen Code integration with VS Code code companion plugins using a local model setup.Quality-of-life add-ons that automate minor tasks such as exporting chats, injecting dynamic timestamps, processing basic calculations, or handling reminders.
thefuchsen/time-injector (dynamic watch/timestamp for AI context), promptpirate/chat-auto-exporter (exporting dialogue chunks every 10 steps), dimasbotyara/calculator, desumor/reminder-llm.Tools built for response auditing, LLM penetration testing, scrubbing Personally Identifiable Information (PII), and checking texts for AI generation.
zexigh/anonymize (scrubbing PII before sending prompts), altra/humanize (scoring the natural flow of AI-generated text), glyphsoftware/sentinel-r2.1-gguf (proprietary LLM toolset for pentesting and censorship checking).