eeg-audit

You validate EEG claims by cross-checking analysis-plan traceability, stats JSON numeric integrity, and report or figure consistency. This process detects post-hoc edits and confirms p-values, effect sizes, and permutation counts match across all pipeline artifact directories.

COBIDAS-MEEG compliance ensures method completeness in EEG research reports. You check it by auditing your end-to-end EEG pipeline outputs, which generates a structured JSON output contract highlighting compliance gaps and missing methodological details.

Yes, you can invoke an external Codex MCP reviewer with no prior context to perform a hostile methods review. This independent LLM reviewer evaluates your bundled audit inputs and generates per-claim verdicts alongside action items for statistical validation.

You detect post-hoc edits by running deterministic mtime integrity checks against your frozen analysis plan and stats generation. This guardrail verifies that the analysis plan was not modified after statistical outputs were generated, ensuring reproducibility.

Yes, the skill validates permutation counts, seeds, and tails by cross-referencing them across the analysis plan, stats JSON outputs, and report text. This ensures cross-file numeric consistency and verifies that reported statistical parameters match the actual computed values.

You should run the audit after your end-to-end EEG analysis stages finish and claims are generated. Running it at this point ensures the tool can effectively cross-check plan traceability, stats JSON integrity, and report consistency, producing actionable fixes in the audit-stage directory.