Public arXiv client — search preprints, fetch paper metadata and abstracts, generate Markdown notes and BibTeX, and download PDFs or source archives. Use for arXiv IDs, preprints, abstracts, citations and full paper artifacts.
Browser automation over Chrome DevTools Protocol (CDP) on :9222. Use to inspect and navigate named tabs, capture compact accessibility or readable-content snapshots, interact with pages, run Google workflows, and take screenshots in the user's real Chrome.
Query local NDJSON/JSONL logs, JSON, CSV, Parquet, and DuckDB files with SQL. Use for filtering logs, aggregation, schema discovery, timelines, counts, error analysis, and ad-hoc analytics over workspace files.
Google Cloud Storage client — list and read objects, upload files/content, copy and delete objects, and manage buckets through the Storage JSON API. Use for gs:// paths, GCS buckets and sharing files through cloud storage.
GitHub REST API client — repositories, issues, pull requests, changed files, reviews, comments and GitHub search. Use for GitHub repos, code/issues/PR search, issue discussions and PR review context. Writes are guarded by explicit confirmation.
Google account toolkit for the uniskill runtime — one OAuth broker plus Gmail, Google Calendar, Docs/Drive, Sheets, Tasks and Workspace Admin Directory, with browser-driven Google web + Flights search. Use when the user asks about reading or sending email, their calendar/events/meetings, Google Docs or Drive files…
Google Places API client — text and nearby search, place details, addresses, phone, website, hours, ratings, reviews and photos. Use to find restaurants, cafes, pharmacies, hospitals, hotels, stores and other real-world locations.
Standalone Hacker News mirror — configured Algolia feeds, normalized stories, feed membership, per-feed cursors and incremental sync runs. Use to sync or browse HN independently. It does not publish to News.
Typed knowledge graph for people, organizations, products and concepts — canonical JSON facts, provenance, relations, validation, person matching and full-text search in the isolated Postgres schema knowledge. Use to look up known entities, contacts, organizations, products, concepts, facts and their sources.
Durable news archive and reader — source-neutral item storage, search, filtering, read state and likes in isolated Postgres schema news. Use to browse or query already stored news. This core plugin intentionally does not fetch RSS, run jobs, summarize articles, or depend on source integrations.
Evidence research over 220M+ peer-reviewed papers through Consensus — ask synthesized research questions, search ranked papers, inspect citations and continue contextual follow-ups. Use for scientific, medical and academic evidence questions.
Durable RSS/Atom feed library — subscriptions, incremental deduplicated loading three times daily, feed state and publication into News with shared one-paragraph summarization.
Standalone Substack producer — subscription catalogue, archive metadata, authenticated full post bodies, per-publication cursors and strict incremental sync through Hyper Browser. It does not publish to News yet.
Use this plugin to create and run durable tasks with one attached agent chat per task. Data is stored in PostgreSQL relation tasks.task; the UI is available at /tasks.
Personal Telegram client over MTProto — find chats by title or username, list recent dialogs, read message history without marking it read, search messages globally or within a chat, find contacts, list folders and group participants, and perform guarded writes.
Google Cloud Text-to-Speech — synthesize text or markdown into OGG, MP3 or WAV with multilingual voices, and list available voices. Use when the user asks to narrate text, read something aloud or generate an audio file.
YouTube Data API and yt-dlp client — search videos/channels/playlists, fetch metadata and statistics, and extract captions as plain-text transcripts. Use for YouTube URLs, video research, channel stats, playlists, subtitles and transcripts.
Live Zulip REST API client for configured instances — list channels, topics and users; read and search messages and DMs; optionally send, edit and mark messages read. Use for Zulip streams, channels, topics, direct messages, full-text search, chat.fhir.org and Health Samurai chat. Credentials stay in 1Password.
Low-level Brave Search API integration. Use only when Brave-specific filters or the raw provider response are required; prefer websearch.search for ordinary public-web discovery.
Read-only Pipedrive CRM client — inspect deals, people, organizations, pipelines and stages, search CRM records, and read email threads attached to deals or contacts. Use for sales pipeline, customer, lead, contact and CRM questions.
Provider-neutral public web search and focused page processing. Use to retrieve ranked links through Brave or Google Browser, then open a selected public page and apply an LLM instruction to its readable content.
★not rated 11 1mo agoA41 tokens
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: