What problem does it solve?
Annotated research documents often contain errors such as reversed relation directions, duplicate ids, over-broad annotation boundaries, or dialect-invalid syntax that break parsing or distort scientific meaning. This Skill audits MIRA annotations in Markdown files and produces actionable findings without modifying the source files.
Core Features & Use Cases
- Semantic and Relation Review: Verifies that claims, evidence, questions, protocols, and requests match their annotated text, and that relations like supports/supported-by have correct direction, existing targets, and textual grounding.
- Multi-Dialect Validation: Checks annotation syntax against Stencila Markdown, Quarto Markdown, MyST Markdown, and plain Markdown conventions, including fences, directive options, and id formats.
- Severity-Ranked Findings: Outputs findings tables with Critical, Major, Minor, or Question severities, locations, explanations, and recommended fixes.
- Use Case: A researcher has a Quarto manuscript with MIRA claim and evidence annotations. Run this Skill to detect a claim that points to itself as supporting evidence and receive a concrete fix recommendation before publication.
Quick Start
Review the MIRA annotations in manuscript.qmd for semantic correctness, relation integrity, and dialect validity, and report findings without editing the file.
