Research and verify multi-source reports with citations
Fan-out research workflow that runs 5 parallel web searches, verifies claims by 3-vote consensus, and synthesizes a cited report.
Why it matters
Produce comprehensive, fact-checked research reports by fanning out web searches across multiple angles, fetching and extracting claims from top sources, adversarially verifying each claim through three-vote consensus, and synthesizing findings into a cited report ranked by confidence.
Outcomes
What it gets done
Decompose research questions into five distinct search angles and execute parallel web searches
Fetch and deduplicate top 15 sources, then extract falsifiable claims from each
Run three-vote adversarial verification on every claim, requiring two-thirds consensus to refute
Merge semantically duplicate findings, rank by confidence, and cite all sources
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/ag-go-in-depth | bash Overview
Go In Depth
Runs the go-in-depth workflow: fans out five parallel web searches, fetches and dedups the top 15 sources, verifies each claim through a 3-vote adversarial process, then synthesizes a cited report. Use it for a deep, multi-source, fact-checked report, not for quick facts - the multi-agent search and verification take significant time.
What it does
Runs the "go-in-depth" workflow: a research harness that fans out multiple web searches, fetches sources, adversarially verifies claims, and synthesizes a cited report. Internally it moves through five phases: Scope (decompose the question into 5 search angles), Search (5 parallel WebSearch agents, one per angle), Fetch (dedup URLs, pull the top 15 sources, extract falsifiable claims), Verify (3-vote adversarial verification per claim, needing 2 of 3 votes to refute and kill a claim), and Synthesize (merge semantic duplicates, rank by confidence, cite sources).
When to use - and when NOT to
Use it when the user wants a deep, multi-source, fact-checked research report on any topic. Before invoking, check whether the question is specific enough to research directly; if it's underspecified (its own example: "what car to buy" without a budget, use case, or region), ask 2-3 clarifying questions to narrow the scope first, then pass the refined question through args with the answers woven in. It is not built for quick facts: multi-agent search, fetching, and 3-vote verification all take significant time, and analyzing 15 full sources is context-intensive, so avoid it for anything that doesn't need that depth.
Inputs and outputs
Takes a query argument describing the research question, for example:
Workflow({ name: "go-in-depth", args: { query: "best hybrid cars under $30k in the US for families" } })
It can also be invoked with no args to run the workflow directly (Workflow({ name: "go-in-depth" })). Output is a synthesized, cited report built from up to 15 deduplicated sources.
Integrations
Runs as a named Workflow (go-in-depth), invoked via scripts/workflow-script.js, and composes WebSearch agents internally for the Search phase.
Who it's for
Anyone who needs a fact-checked, multi-source report rather than a single quick answer - including fact-checking a specific claim (its own example: efficacy of intermittent fasting for long-term weight loss) or a deep dive into a technical concept (its own example: how transformers handle positional encoding).
Source README
Go In Depth
Overview
Go in depth harness - fan-out web searches, fetch sources, adversarially verify claims, synthesize a cited report. Run the "go-in-depth" workflow.
When to Use
When the user wants a deep, multi-source, fact-checked research report on any topic. BEFORE invoking, check if the question is specific enough to research directly - if underspecified (e.g., "what car to buy" without budget/use-case/region), ask 2-3 clarifying questions to narrow scope. Then pass the refined question as args, weaving the answers in.
How It Works
Phases:
- Scope: Decompose question (from args) into 5 search angles
- Search: 5 parallel WebSearch agents, one per angle
- Fetch: URL-dedup, fetch top 15 sources, extract falsifiable claims
- Verify: 3-vote adversarial verification per claim (need 2/3 refutes to kill)
- Synthesize: Merge semantic dupes, rank by confidence, cite sources
Examples
Example 1: Run go-in-depth workflow
Workflow({ name: "go-in-depth" })
Example 2: Research with refined question
Workflow({ name: "go-in-depth", args: { query: "best hybrid cars under $30k in the US for families" } })
Example 3: Deep dive into a technical concept
Workflow({ name: "go-in-depth", args: { query: "how does the transformer architecture handle positional encoding?" } })
Example 4: Fact-checking a medical claim
Workflow({ name: "go-in-depth", args: { query: "efficacy of intermittent fasting for long-term weight loss in adults" } })
Workflow Script
Limitations
- Slow execution: Multi-agent searches, fetching, and 3-vote verification take significant time. Not for quick facts.
- Context intensive: Analyzing 15 full sources uses large context limits.
- Synthesis risks: May struggle if source material is weak or equally conflicting.
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.