open-deep-research

name	open-deep-research
description	Use this skill when the user wants a scholarly or clinical literature review, evidence scan, citation-graph exploration, or OpenAlex-based source discovery across many papers. It installs and uses the `open-deep-research` CLI from PyPI, runs deep research through OpenAlex, expands through references and citations, and returns a Markdown report plus inspectable trace files.

Open Deep Research

Run broad literature discovery when a local corpus is not enough.

Workflow

1. Ensure the CLI is installed or upgraded:

bash <skill-directory>/scripts/install_or_upgrade.sh

2. Choose the output mode:

• Use --format report when the user wants a readable answer immediately.
• Use --format paths when you expect to inspect report.md, papers.json, or trace.json afterward.
• Use --no-llm only when no LLM provider is configured or when you want a retrieval-only baseline.

3. Run the research job with stdin, not shell quoting:

printf '%s' "$QUESTION" | bash <skill-directory>/scripts/run_report.sh

For inspectable artifact paths:

printf '%s' "$QUESTION" | bash <skill-directory>/scripts/run_paths.sh

Defaults

• Prefer OpenAlex-backed discovery for paper finding and citation expansion.
• Prefer the PyPI package open-deep-research-cli rather than editing the repo in place.
• Prefer report.md for direct user-facing synthesis.
• Prefer reading trace.json when you need to justify why papers were selected.

Clinical Use

• Use this skill for evidence gathering, not for autonomous clinical decisions.
• When the user asks a medical question, present the result as a literature scan or evidence summary, not as personal medical advice.
• State clearly when the output is based on abstracts or partial open-access text rather than full articles.

Outputs

The CLI writes:

• report.md
• papers.json
• trace.json

Read references/commands.md when you need exact command variants or output behavior.