additional_features
["Strict Factual Mode (evidence-only; strict numerical and terminology fidelity)", "StoryFlow evaluation module (narrative spine, coherence scoring, breakpoints, bridge text generation)", "Checklist-driven methodology audit engine (AAPOR, weighting, variance estimation, measurement equivalence, reproducibility, uncertainty visualization)"]
example_commands
["Evaluate the attached report using the 6-axis framework. Include a full methodology audit (AAPOR, weighting, variance estimation, DEFF, missingness/imputation, scale validity, reproducibility).", "Run a representative-chapter review and include Story Mode outputs (Narrative_Spine, Breakpoints, Bridge_Texts).", "Identify where the narrative breaks between tables and interpretations, list Breakpoints, and draft Bridge_Texts in the original tone.", "Check whether time-series indicators are compar
gpt_id
g-689b6ef185d881918d1f4efcc3b12930
ideal_use_cases
["Audit survey reports for response rates (AAPOR), sampling, weighting, variance estimation, and DEFF compliance", "Identify logical gaps between results, interpretation, and recommendations and repair narrative flow", "Detect risks in time-series or subgroup comparisons due to definition changes or measurement non-equivalence", "Ensure uncertainty is properly communicated in tables and figures (CIs, error bars, DEFF)", "Generate a final pre-submission quality evaluation and revision guide"]
limitations
["All judgments are strictly based on the provided source text and attachments (Strict Factual Mode)", "No speculation or inference beyond the evidence (zero-inference rule)", "Does not answer general user questions; only produces evaluation outputs", "Will not invent, adjust, or retrofit numbers, definitions, or methods absent from the source", "Marks items as 'insufficient evidence' when documentation is lacking"]
target_users
["Policy and research report quality assurance owners", "Survey research practitioners (design, fielding, analysis)", "Authors and reviewers of public, private, or academic research reports", "Editors, peer reviewers, and internal audit teams"]