Gemini researches Stonehenge into a local report
Hutchison reports Gemini producing a cited Stonehenge research report as a local Markdown file in about 20 minutes through his terminal extension.
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Research [historical question] using primary and academic sources. Track the long-running job, save a cited Markdown report to [folder], distinguish uncertain conclusions, and list source checks I should perform before using it.
Reach people exploring AI agents
See placements →What happened
Hutchison reports Gemini producing a cited Stonehenge research report as a local Markdown file in about 20 minutes through his terminal extension.
Reported by Allen HutchisonHow the agent did it
- Install the Gemini CLI Deep Research extension and provide an API key.
- Start a focused historical research question from the terminal.
- Let the research agent plan, query sources and synthesise findings.
Where a person stepped in
Hutchison chooses the research question and follow-up, reads the report and uses its citations. The post does not establish independent historical fact-checking.
Allen Hutchison wanted research results to appear directly in his file-based writing workflow. After losing an earlier browser report, he built a Gemini CLI extension exposing Gemini Deep Research through MCP. His first concrete trial asked about the scientific evidence for Stonehenge construction after reading Ken Follett’s novel Circle of Days.
The research job planned its investigation, queried sources and synthesised a report. The CLI checked whether the long-running job had completed rather than requiring Hutchison to manage it in another browser tab. He reports that a Markdown file appeared in the local filesystem about twenty minutes later, containing citations to academic material, isotope analysis and archaeological evidence. A reference to the Bell Beaker transition prompted a second investigation using the same workflow.
The practical detail is to preserve a long research task as a tracked job with local state, then save the artifact where the writer already works. The extension also supports indexed local source collections, although the Stonehenge trial was the reported example. Hutchison describes daily use but does not independently assess the report’s historical accuracy. Readers can copy the output-management pattern while still checking important citations and distinguishing the agent’s synthesis from established archaeological conclusions.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Blog · Allen Hutchison
Frequently asked questions
What did the first research job investigate?
Scientific evidence about how Stonehenge was constructed and who built it, prompted by the author’s reading.
Where did the report appear?
A cited Markdown file appeared in the local filesystem after roughly twenty minutes, according to Hutchison.
Was its accuracy independently verified?
The account reports a useful cited synthesis, but no independent evaluation of every archaeological claim.
More details
Remaining steps
- Check job completion and save the cited Markdown report locally.
- Use a finding in that report to start a follow-up investigation.
About this account
Gemini carries out a long research investigation and the terminal client polls completion, delivering a persistent local artifact instead of a transient browser response.
Firsthand account by Allen Hutchison; original publication date verified against Primary page publication date. Outcomes are the author’s reports, not independently reproduced. Suggested prompt written by Agents In Use; not the author’s original prompt.
Credit: Allen Hutchison
Source published 20 Dec 2025
Source checked 14 Oct 2026
Tools and topics
Gemini CLI · Gemini Deep Research · MCP · Markdown
Historical research · Cited reports · Local artifacts
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.