Sopaper Evidence: Evidence-First Research Workflow
Sopaper Evidence is a rigorous, structured methodology designed for researchers and academic writers who need to ground their work in verifiable facts rather than assumptions or fabrications. This skill operates on a foundational principle: no claim without evidence.
Core Functionality
The workflow systematically transforms vague research ideas into citation-backed, audit-ready evidence packs. It guides users through six integrated stages: (1) scoping the research question, (2) searching for prior work and datasets, (3) verifying and classifying sources, (4) extracting structured evidence, (5) building comprehensive evidence maps, and (6) only then supporting actual writing tasks.
Key Strengths
Anti-Hallucination Design: The skill enforces hard rules against fabricating papers, authors, venues, citations, or numerical results—a critical safeguard in an era where AI systems frequently generate plausible-sounding but false academic references.
Source Hierarchy: It implements a three-tier priority system (user artifacts → primary sources → secondary summaries) ensuring maximum reliability.
Structured Outputs: Multiple templated deliverables (claim-evidence maps, related-work matrices, experiment gap reports) make the evidence audit trail transparent and reusable.
Automation Support: 13 bundled Python scripts handle pipeline stages from search plan generation to gap triage, reducing manual overhead while maintaining rigor.
Notable Limitations
Labor-Intensive: The thoroughness that prevents errors also demands significant effort. Users seeking quick drafts may find the workflow constraining.
Dependency on Accessible Sources: Effectiveness depends on the availability of verifiable external sources. Niche or proprietary domains may leave persistent evidence gaps.
No Writing Creativity: This skill deliberately defers all generative writing until evidence is complete—creative writers may chafe at the procedural discipline.
Ideal Users
- Graduate students preparing theses or publications
- Research teams requiring reproducible literature reviews
- AI/ML practitioners in embodied robotics (with OpenClaw-specific extensions)
- Anyone burned by hallucinated citations from general-purpose AI tools
Risk Assessment
Primary Risk: Evidence gaps may halt progress unexpectedly. Users must be prepared to pivot research questions or conduct original experiments when verification fails.
Secondary Risk: Over-reliance on automation scripts without manual source review could propagate verification errors.