comprehensive-researcher — Claude Code subagent: sourced research
A Claude subagent that turns a topic into research questions, gathers sources and writes a cited report. Useful when you need an overview, not a guess.
npx claude-code-templates@latest --agent podcast-creator-team/comprehensive-researcherWhat comprehensive-researcher does
The subagent runs in steps. It breaks a topic into 5–8 specific research questions, then looks for at least 3–5 sources per question: academic papers, government and institutional reports, established news outlets, expert commentary. Each source gets checked for credibility and recency, and the agent notes where authors agree and where they clash.
What comes back is a structured report: an executive summary of key findings, an introduction setting the scope, a main body organised by question or theme, inline citations next to claims, and a full bibliography. Cross-checking is its own step — facts get verified across sources, controversial points are shown from several sides, and when reliable material is thin the agent says so instead of filling the gap.
Under the hood it is a single agent file describing the process, with four tools declared: Read, Write, Edit, WebSearch. So web search access is what makes the sourcing step work, and file write access is what lets it save the report. Run it somewhere without web search and that stage simply does not happen.
What it helps with
- break an unfamiliar topic into answerable questions
- produce a report with citations
- verify facts across sources and surface disagreements
- separate confirmed findings from opinion
- get several viewpoints on a contested topic
Example prompts for the agent
- research CI/CD options for small teams and write a cited report
- break down vector databases into questions, sources and conclusions
- survey API caching approaches and flag where experts disagree
How to install comprehensive-researcher
1claude-code-templates
npx claude-code-templates@latest --agent podcast-creator-team/comprehensive-researcher2Manually: put the files in place
.claude/agents/comprehensive-researcher.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/comprehensive-researcher.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/comprehensive-researcher.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about comprehensive-researcher
How do I install comprehensive-researcher in Claude?
Use the templates CLI: npx claude-code-templates@latest --agent podcast-creator-team/comprehensive-researcher. The agent file then sits at .claude/agents/comprehensive-researcher.md.
What does comprehensive-researcher do?
It decomposes a topic into research questions, searches for sources, cross-checks the facts and compiles a structured report with citations.
Which tools does the subagent need?
It declares Read, Write, Edit and WebSearch. Without web search access the sourcing step cannot run.
