claude-scientific-writer

Ask about a topic and last30days searches recent discussions across Reddit, X, YouTube, Hacker News, Polymarket, GitHub, and the wider web, then synthesizes cited findings using signals such as votes, comments, or market odds. It can fall back to web-only research without keys, while richer sources require your own API credentials, browser sessions, or helper tools, and it runs in many Agent Skills hosts or as a Claude Desktop MCP bundle.

Cairn records trust ratings for web pages, search results, MCP tools, and network-bound commands so later agents can check how those resources performed for other reviewers. In Claude Code, post-tool hooks can rate calls in the background; Desktop and web installs expose explicit scoring tools instead. Each rating consumes either Claude subscription quota or Anthropic API spend, and rating data is sent to the Cairn service, so cadence and privacy controls matter.

Ask about a container tradelane, port, liner service, carrier network, or freight-rate movement, and the skill searches public alliance documents, carrier releases, port statistics, indices, and trade reporting with citations. It bundles a research method rather than a dataset and excludes air, inland, bulk, and tanker transport. Exact rotations, transit times, cut-offs, and rates still need confirmation from the carrier or forwarder before a commercial decision.

This small Claude Code plugin library currently offers two manual workflows. The ETL skill stages raw research material before transformation, while Tournament dispatches multiple parallel drafters, ranks their outputs with a decisive judge, and can synthesize the strongest result; automatic model invocation is disabled so people choose when the extra process is worth using.