Research Router
Route research requests to the right source class, then return source-backed answers with confidence and gaps.
Routing policy
- Prefer live data for open-web questions.
- Pick the smallest route set that can answer the question.
- If evidence is thin, add a second source class for corroboration.
- Keep source provenance explicit in final output.
Route map
Code/docs research
- Use
ghfor GitHub repository inspection and code search. - Use
deepwikifor repository docs/Q&A overowner/repo. - Use
context7for up-to-date library/API docs and examples. - Use
grep-appfor public OSS code usage patterns.
Web/live research
- Use
brave-searchfor fast scoping and recency checks. - Use
exa-searchfor deeper multi-source synthesis and richer retrieval. - Escalate from Brave to Exa when coverage or quality is weak.
Sentiment/discussion research
- Use
redditfor community sentiment, discussion trends, and user/topic signals. - Corroborate high-impact claims with one non-Reddit source class when feasible.
Knowledge-base research
- Use
notebooklmfor user-owned notebook sources and internal knowledge corpora.
Exclusions
- Do not route to
summarizefrom this skill.
Router workflow
- Classify request: code/docs, web/live, sentiment, or notebook KB.
- Select primary route from the map above.
- Execute route tools and collect source evidence.
- Add corroboration route when confidence is low, sources conflict, or claim impact is high.
- Return answer with provenance, confidence, and explicit gaps.
Output contract
- Answer: concise result.
- Sources used: tool + why it was selected.
- Confidence: high/medium/low with reason.
- Gaps: unresolved uncertainty or missing evidence.
- Next query: best follow-up when blocked.
Escalation rules
- If library/API details are unclear, force
context7. - If repo-specific behavior is unclear, force
deepwikiand/orgh. - If web evidence is stale or shallow, escalate from
brave-searchtoexa-search. - If sentiment claim drives decisions, include
redditplus one non-Reddit corroboration source.
