Terminal
Terminal/Skill
Runs a web research pipeline that searches for sources, fetches them as clean markdown and summarises them (needs Python 3.8 or newer).
By MiniMax
Works with
npx skills add minimax-ai/minimax-code-plugins --skill web-researchEnd-to-end web research pipeline: search → fetch with a free-first cascade → summarise.
Search (find candidate URLs): Tavily /search (keyless by default). Defuddle is not a search engine and is not used here.
Fetch (one URL → clean markdown): cascade in this order, stop at first success.
command -v defuddle returns non-zero. Do not install defuddle automatically; tell the user once if it is missing./extract (keyless, free but rate-limited) — fallback when defuddle is missing or fails on a URL. Header X-Tavily-Access-Mode: keyless is sent by scripts/tavily_extract.py.<!-- needs mcode browser: <url> --> marker; the model sees the marker and re-fetches that URL with the browser tool.Each output file is named NN-<slug>.md and prefixed by a tag noting which layer produced it ([defuddle], [tavily extract], or [browser required]).
If the user did not specify constraints, run the full pipeline in one shot:
python3 scripts/research.py "<query>" 5 ./research-output
The script returns non-zero if Tavily search itself failed; partial URL failures are surfaced as [browser required] markers inside the output directory, not as script errors.
Search for URLs only:
python3 scripts/tavily_search.py --query "<query>" --max-results 5 --format brave
Output is JSON {query, answer, results:[{title,url,snippet}]}.
Fetch one URL with the cascade (the script picks the layer based on availability):
# Preferred (defuddle must be installed):
defuddle parse "<url>" --md -o "<out>.md"
# Fallback (Tavily keyless extract):
python3 scripts/tavily_extract.py --url "<url>" --output "<out>.md"
# Last resort (model-driven):
# use the mcode browser tool on <url>
Combine the resulting markdown files into a short summary at the end.
TAVILY_API_KEY for higher limits, and continue. The script does not auto-retry.research.py output (to stderr), then continue to Tavily extract for every URL. Do not install defuddle.<!-- needs mcode browser: <url> --> marker file and continue. The model re-fetches with the browser tool only if the user cares about that URL.python3 scripts/research.py "向量数据库 最新进展" 5 ./research-outputpython3 scripts/research.py "Rust async runtime comparison" 3 ./research-outputscripts/research.py — the one-shot pipeline (search → cascade fetch → tag outputs). Python, no bash dependency.scripts/tavily_search.py — Tavily /search in keyless mode. Single Python file, stdlib only.scripts/tavily_extract.py — Tavily /extract in keyless mode. Single Python file, stdlib only.No native binaries. No symlinks. No secrets in tree. TAVILY_API_KEY is read from the environment only and never logged.