Research deep

--- description: Vault-first deep research:scans the vault, fills gaps (using Perplexity and Grok when a key is provided, and free, key-less sources otherwise), synthesizes a delta, then propagates updates across people, projects, and ideas via/obsidian-save category: research triggers_en : ["deep research", "thorough research", "vault-first research", "research gaps"] --- Use the obsidian-second-brainskill . Execute/research-deep [topic] : 1. Determine the topic based on the user’s argument. If no topic is provided, ask: “What topic for deep research?” 2. Run the Python command from therepo root (~/Projects/personal/obsidian-second-brain/ ): ``bash uv run -m scripts.research.research_deep "<topic>" ` The script auto-selects its mode: if PERPLEXITYAPIKEY is set it runs the paid pipeline below; otherwise it falls back to free, key-less sources. Pass --free to force free mode, or --academic (free mode only) to restrict to scholarly sources. Phase 1 (vault scan) is identical in both modes, so OBSIDIAN_VAULT_PATH must be set either way. 3. **Paid mode** - the script runs a 4-phase pipeline and finishes the work itself: - **Phase 1** - vault scan: finds existing notes mentioning the topic (the baseline). - **Phase 2** - gap analysis: Perplexity sonar-pro identifies what's missing/stale and emits 3-5 targeted queries (each tagged web or x). - **Phase 3** - gap-fill: runs each query via Perplexity (web) or Grok+Live Search (X discourse). - **Phase 4** - synthesis: Perplexity produces a delta report, the script saves it to Research/Deep/YYYY-MM-DD - <slug>.md, then emits a JSON payload between <<<RESEARCHDEEPPROPAGATION_PAYLOAD >>> markers. Show the synthesis body verbatim, then do the propagation step (step 5). 4. **Free mode** - the script does Phase 1 (vault scan) plus free-source aggregation and prints a JSON block with "mode": "free-sources-deep", containing vaultbaselinenotes (path, score, excerpt of what the vault already knew), sources (fresh external results), stats, warnings, and an instruction. YOU are the synthesizer: - Read the baseline excerpts and the source results. If stats.success is false (fewer than 3 sources returned), flag the thin coverage in Open Questions - do not pad. - Produce a delta with exactly these sections: What's New Since Vault Baseline, What's Confirmed, Contradictions / Updates Needed (name the [[vault path]]), Synthesis, Recommended Vault Updates, Open Questions. Every external claim carries a recency marker and source domain; every vault reference uses [[wikilinks]]. Never invent facts to fill a section. - Save it yourself to Research/Deep/YYYY-MM-DD - <slug>.md per references/ai-first-rules.md (preamble; frontmatter with type: research-deep, ai-first: true, vault-baseline-notes, and a sources list of every result URL verbatim). - Show the synthesis to the user, then do the propagation step (step 5). 5. **Propagation (both modes):** - In paid mode, parse the JSON payload; in free mode, use the note you just wrote and its synthesis. - Treat the synthesis body as the "conversation context" input to /obsidian-save. - Run the standard /obsidian-save ` flow: spawn parallel subagents (People, Projects, Tasks, Decisions, Ideas) and update vault notes according to the synthesis's "Recommended Vault Updates" bullet points. - Apply the AI-first vault rule to every note created or updated (preamble, frontmatter, recency markers, wikilinks, sources). - Link the new research note from today’s daily note. - Then report back a clean list: “Updated [[X]], created [[Y]], linked [[Z]] from today’s daily note.” 6. Plain English triggers: "do deep research on [topic]," "research [topic] properly," "vault-aware research on [topic]," "research and update the vault on [topic]." 7. If any source or phase fails (Grok unavailable in paid mode, or a free source times out), the run continues with what it has and flags the gap in the synthesis. Display partial results:don’t fail silently. The graceful degradation rule: a partial synthesis is better than no synthesis. 8. Cost note: Paid mode makes multiple calls to theAPI (Perplexity + Grok), typically costing $0.20:$0.80 per day.