Deep Research — n8n

Turn a fuzzy question into a sourced research brief with open questions and next checks — not a final verdict. (n8n workflow recipe (Manual Trigger → LLM → output).)

Id: n8n/deep-research · Slug: deep-research · Status: Untested · Untested on this machine — recipe reviewed for safety (no secrets, no destructive commands, no auto-spend). Mark tested after you run it locally.

What is this

Workflow for n8n. Turn a fuzzy question into a sourced research brief with open questions and next checks — not a final verdict. (n8n workflow recipe (Manual Trigger → LLM → output).)

How this runtime fits

This n8n implementation is primarily an AI workflow / agentic workflow for the parent AI Team. Call it a multi-agent workflow only when the graph actually splits agent-like steps.

Parent AI Team: deep-research · What is an AI Team?

What it does

Turn a fuzzy question into a sourced research brief with open questions and next checks — not a final verdict. (n8n workflow recipe (Manual Trigger → LLM → output).)

Who for

Builders working on Research jobs who can run n8n themselves.

Need to run

You will run this locally or on infra you control.
No production credentials in prompts or committed files.
n8n instance you control
Optional local LLM HTTP endpoint

How to use

In n8n: Import from File → workflow.json (or recreate nodes from the sketch).
Add your local LLM credential/HTTP node in place of the placeholder.
Paste the full system prompt into the LLM node.
Execute once with Example in as input. Inspect output; no schedule yet.

Limitations

Not a substitute for professional legal, medical, or investment advice.
Outputs can be wrong; human review required before publish or spend.
Stage-1 Library items are free recipes — marketplace ready-to-use teams remain separate.
Do not attach billing cloud LLM nodes without a human budget check.

Test status

Status: Untested · Untested on this machine — recipe reviewed for safety (no secrets, no destructive commands, no auto-spend). Mark tested after you run it locally.

Files

workflow.json — Importable n8n workflow sketch
README.md — Safe import notes

Full prompt / config / code

# n8n Recipe — Deep Research

## Full system prompt (put in your LLM node)
You are a careful research desk. Goal: produce a brief, not a verdict.

RULES:
- Use only facts present in the user paste or clearly marked as [UNVERIFIED].
- Separate: Known / Inferred / Unknown / Next checks.
- Cite sources as the user labeled them (URL, doc name, date). Do not invent citations.
- No spending, no shell commands, no credential requests.
- Stop after the brief. Ask one clarifying question only if blocking.

OUTPUT:
1) One-sentence scope
2) Findings (bullets)
3) Open questions
4) Next checks (max 5)
5) Confidence: low|medium|high + why

## workflow.json
```json
{
  "name": "BotShelf Library \u2014 Deep Research",
  "nodes": [
    {
      "parameters": {},
      "id": "1",
      "name": "Manual Trigger",
      "type": "n8n-nodes-base.manualTrigger",
      "typeVersion": 1,
      "position": [
        0,
        0
      ]
    },
    {
      "parameters": {
        "values": {
          "string": [
            {
              "name": "system",
              "value": "You are a careful research desk. Goal: produce a brief, not a verdict.\n\nRULES:\n- Use only facts present in the user paste or clearly marked as [UNVERIFIED].\n- Separate: Known / Inferred / Unknown / Next checks.\n- Cite sources as the user labeled them (URL, doc name, date). Do not invent citations.\n- No spending, no shell commands, no credential requests.\n- Stop after the brief. Ask one clarifying question only if blocking.\n\nOUTPUT:\n1) One-sentence scope\n2) Findings (bullets)\n3) Open questions\n4)\u2026"
            },
            {
              "name": "user_input",
              "value": "={{$json.input}}"
            }
          ]
        }
      },
      "id": "2",
      "name": "Prepare Prompt",
      "type": "n8n-nodes-base.set",
      "typeVersion": 3,
      "position": [
        260,
        0
      ]
    },
    {
      "parameters": {
        "notice": "Attach your own local LLM node (Ollama/LM Studio HTTP). Do not enable billing cloud nodes without a human budget check."
      },
      "id": "3",
      "name": "LLM Placeholder",
      "type": "n8n-nodes-base.noOp",
      "typeVersion": 1,
      "position": [
        520,
        0
      ]
    }
  ],
  "connections": {
    "Manual Trigger": {
      "main": [
        [
          {
            "node": "Prepare Prompt",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prepare Prompt": {
      "main": [
        [
          {
            "node": "LLM Placeholder",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  },
  "meta": {
    "botshelf_job": "deep-research",
    "safety": "manual-trigger-only"
  }
}
```

## README safety
- Trigger is Manual only (no cron spend loops in stage 1).
- Replace LLM Placeholder with a local HTTP node to Ollama/LM Studio when ready.
- Do not store API keys in the workflow JSON committed to git; use n8n credentials store.
- No shell nodes that delete files or push to production.

Example in

Question: Why did our signup conversion drop last week?
Sources:
- [A] analytics export 2026-09-01..09-07 (CSV summary pasted)
- [B] changelog: checkout button copy tweak on 09-03
Ask for a brief with next checks.

Example out

Scope: Explain likely drivers of signup conversion drop using only pasted analytics + changelog.
Findings:
- Drop concentrated on mobile checkout step (from A)
- Timing aligns with 09-03 copy tweak (B)
Open questions:
- Was A/B still running?
Next checks:
1) Diff mobile CTA copy pre/post
2) Error rate on payment iframe
Confidence: medium — single-week window.

Model / runtime notes

Stage-1 recipe is manual-trigger only. Add schedules only after you confirm cost controls.

Canonical Team

Parent: deep-research · implementation_id: n8n/deep-research

GitHub source

teams/deep-research/n8n · commit eb87e8e049fdf9908e438305ff827c7b617505b5 · license: free-use-at-own-risk

Structural check: PASS · env: structural (structural only — does not set Verified)

Related Library items

Related marketplace Team pages

Soft thematic links only — not identity merges. Marketplace Teams ≠ Library AI Team records.

Research Desk (marketplace Team page — separate entity)
Source Desk (marketplace Team page — separate entity)