expert-interviewer

Runs retrieval-grounded simulated expert interviews per STORM perspective with source-backed answers.

2|1|Updated Mar 7, 2026
One-click install
npx skills add https://github.com/AutoByteus/autobyteus-agents --skill expert-interviewer-autobyteus
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: expert-interviewer
Source: https://github.com/AutoByteus/autobyteus-agents/tree/main/agent-teams/storm-team/agents/expert-interviewer/skills/expert-interviewer
Command: npx skills add https://github.com/AutoByteus/autobyteus-agents --skill expert-interviewer-autobyteus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It automates the knowledge-curation stage of STORM-style article production by simulating expert interviews for each perspective, ensuring every factual answer is grounded in retrieved sources rather than unsupported snippets. ## Core Features & Use Cases - Perspective-Guided Questioning: Reads perspectives.json and runs a question thread per active perspective, asking seed questions and follow-ups when answers reveal new gaps. - Source-Backed Answers: Retrieves and reads sources, records them in source-index.json, and attaches source IDs to every factual answer. - Structured Handoff: Writes interview-notes.md from templates and produces a file-backed handoff package for the outline_architect stage. - Use Case: In a multi-agent STORM pipeline researching a topic, this skill generates the evidence package of questions, answers, and indexed sources that downstream outline and drafting agents consume. ## Quick Start Run the expert-interviewer skill on the topic brief and perspectives file to produce interview notes and a source index for the outline architect.

Frequently Asked Questions about expert-interviewer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run simulated expert interviews for STORM article generation?▼

Provide a topic-brief.md and perspectives.json, then the skill runs a question thread per active perspective: seed questions, source retrieval, source-backed answers, and follow-up questions. Results are written to interview-notes.md and source-index.json.

What inputs does the expert interviewer stage need?▼

It requires two upstream artifacts: topic-brief.md describing the subject and perspectives.json listing the active perspectives. Both must exist in the project workspace before the interview threads begin.

How are sources tracked during expert interviews?▼

Every retrieved source is registered in source-index.json with title, URL, publisher, date, source type, and reliability notes. Each factual answer in interview-notes.md references these source IDs, and search snippets are never accepted as final evidence.

What happens after the expert interview stage finishes?▼

The skill writes a handoff file at handoffs/expert-interviewer-to-outline-architect.md listing artifact paths, changes, open questions, and source-quality concerns, then sends the full package to the outline_architect agent.

What are the limitations of simulated expert interviews?▼

Answer quality depends entirely on retrievable sources, so niche topics may leave unresolved questions, which are tracked explicitly. The skill also guards against one perspective dominating the evidence package but cannot create evidence where none exists.