research-pipeline

To automate an end-to-end literature review pipeline, use a fixed five-stage chain consisting of researcher, synthesizer, devils-advocate, fact-checker, and editor. This structure applies standardized handoffs and quality gates for reproducible document generation.

Adversarial review in research editing is a quality gate stage where a devils-advocate role critiques synthesized findings. This process challenges arguments and identifies weaknesses before the document passes to the fact-checker and editor for final publication-ready output.

Generate publication-ready documents conforming to CLAUDE.md style by routing research through an orchestrated pipeline. The final editor stage enforces CLAUDE.md citation practices and formatting standards to deliver a polished, structured output.

Yes, you can recover an interrupted research pipeline. The orchestration implements resilient planning and tracking with progress checkpoints, allowing you to resume the workflow from the last completed stage without losing your research data.

Yes, the research pipeline supports both comprehensive and focused literature reviews. When initializing a new pipeline with a defined topic and scope, you select between these modes to match the required depth and breadth of your research output.