diillson/chatcli/web-research

Web Research

Native, keyless deep research using ChatCLI's own tools — @websearch to discover sources and @webfetch to read them. No third-party paid search middleman required; the configured search chain handles providers.

Method (search → read → synthesize)

1. Decompose the question into 2–4 sub-queries (different angles/keywords).
2. Search each with @websearch; collect the most relevant, recent, authoritative URLs.
3. Read the top sources with @webfetch — don't answer from snippets alone; open the pages.
4. Cross-check any non-trivial claim against at least two independent sources. Flag conflicts.
5. Synthesize a structured answer with inline citations (URL or source name per claim).

Quality bar

• Prefer primary sources (official docs, papers, vendor pages) over aggregators.
• Note publication dates; for "latest"/"current" questions, prefer the most recent.
• Distinguish fact from opinion/marketing. If sources disagree, say so and show both.
• If the web doesn't clearly answer it, say what's uncertain rather than guessing.

Output

End with a short Sources list (title + URL). Keep the body tight; lead with the answer, then the supporting detail.

When to escalate

For long-running monitoring ("watch this topic and tell me when X changes"), pair this skill with @scheduler (periodic re-run) and @send (notify the user on their channel).