Interview-first — clarifying questions before research starts
PASS23sprompt: enartifact: en
Prompt (verbatim)
Using the bioresearcher-deep-research skill, run a deep research report on CAR-T therapy safety in solid tumors.
What happened
- No no-interview prefix: the skill must ask its clarifying questions before researching.
- Even under non-interactive `opencode run --auto`, the agent posted ONE batch of questions and started NO research.
- Objective checks: final answer contains '?'; contains no 'final_report'; none of article_search / trial_search / gene_get was called.
Evidence
Objective checks
checks passed: 4/4
| Fan-out mode | n/a — no research started (the interview turn comes first) |
|---|---|
| Provenance | 3380cc6 · opencode 1.18.29 |
Replay
node demos/run-demo.mjs --only agent-interview-en --publish
Manual-run only — this replay spends LLM tokens (real opencode CLI + network).
In the repository: README · transcript · result.json · artifact directory (permalink)