How to use Claude to build a resource disruption study
Claude Code orchestrated resource- and country-specific research agents, assembled sourced deep dives, and supported a queryable research project about interconnected disruptions.
news.ycombinator.com
Prompt to try
Editorial adaptation, not the source author’s exact prompt. Replace bracketed inputs and review tool permissions first.
Build a sourced study of [event] focused on [resources] and [countries]. Assign separate research agents to each area, produce self-contained deep dives using [source set and date range], and examine cross-domain disruption cascades. Document file relationships and prepare the corpus for [query interface].
Reach people exploring AI agents
See placements →What happened
The author assembled research organized into 76 interconnected files and made it accessible through NotebookLM, a browsable site, and a source repository.
Reported by hrishirc on Hacker NewsHow the agent did it
- Deploy parallel research agents for each resource and country.
- Produce self-contained deep dives with sourced data.
- Use cascade models to identify compounding disruptions across domains.
Where a person stepped in
The post does not describe human source-checking or independent validation of the study’s conclusions.
A person set the goal and reviewed the result.
The author used Claude Code as an orchestration layer for research into resource disruptions associated with the 2026 Iran War. Rather than starting with political commentary, they describe organizing the analysis around physical constraints: production, shipping routes, and consequences when supplies go offline.
Parallel research agents were deployed for individual resources and countries. Each produced a self-contained deep dive with sourced data. Cascade models then identified how disruptions could compound across domains simultaneously. The author says the research drew on major news outlets, think tanks, and government reports from February and March 2026.
The resulting material was organized using an “Agentic Brain” knowledge architecture. A CLAUDE.md file documented how 76 files were organized and interconnected. The author made the research available through NotebookLM, a browsable site, and a source repository, allowing readers to chat with or inspect it.
The post presents conclusions about chips, food, energy, and insurance as findings of this research. It does not describe independent validation of those conclusions or a human source-checking process. The supported outcome here is the assembled, accessible research project, rather than confirmation of its forecasts.
Who it's for: Researchers exploring interconnected resource constraints who want an organized corpus they can inspect and query.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · hrishirc
Frequently asked questions
How was the research divided among agents?
The author deployed parallel research agents by resource and country. Each produced a self-contained deep dive with sourced data, followed by cascade models examining compounding disruptions.
How could readers access the resulting research?
The author offered chat through NotebookLM, a browsable site, and a source repository. A CLAUDE.md file documented the organization and connections of 76 files.
More details
Remaining steps
- Organize and interconnect the research files in a documented architecture.
- Make the research available through NotebookLM, a site, and a repository.
About this account
Claude Code orchestrated parallel research agents, with cascade models connecting findings across resource domains.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=47502843. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: hrishirc on Hacker News
Source published 24 Mar 2026
Source checked 14 Oct 2026
Tools and topics
Claude Code · NotebookLM · CLAUDE.md · GitHub
Research agents · Resource constraints · Knowledge architecture
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.