abstract-structure
Remove domain surface details to expose transferable relational/mechanistic structure at a chosen abstraction level.
Audit train/test leakage, memorization artifacts, temporal leakage, and disclosure pathways that can invalidate evaluation.
Claude Code
$ npx -y skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination --agent claude-code
Codex
$ npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination -a codex -y
Cursor
$ npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination -a cursor -y
Gemini CLI
$ npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination -a gemini-cli -y
GitHub Copilot
$ npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination -a github-copilot -y
OpenCode
$ npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination -a opencode -y
Windsurf
$ npx skills add yogsoth-ai/de-anthropocentric-research-engine --skill audit-data-contamination -a windsurf -y
name: audit-data-contamination description: "Audit train/test leakage, memorization artifacts, temporal leakage, and disclosure pathways that can invalidate evaluation."
Audit train/test leakage, memorization artifacts, temporal leakage, and disclosure pathways that can invalidate evaluation.
required: [dataset_partitions, training_sources, evaluation_records] optional: [release_history, deduplication_keys, access_logs] constraints: [each contamination claim requires a matched record, pathway, and comparison basis]
1. Define partition, temporal, entity, and disclosure boundaries. 2. Match evaluation items against training and public-source records using declared keys. 3. Test suspected leakage pathways and compare affected with unaffected cases. 4. Classify confirmed, plausible, and unassessed contamination with impact notes.
produces: [contamination_matches, pathway_map, affected_case_set, impact_assessment, audit_uncertainties] delta_fields: [findings, evidence_updates, uncertainties, decisions, open_questions]
Do not call lexical overlap contamination without a disclosure pathway, and do not infer clean separation from missing metadata.
A research skill graph for AI-native science. 271 markdown files. No runtime, no dependencies, no build step.
Remove domain surface details to expose transferable relational/mechanistic structure at a chosen abstraction level.
Evaluate competing arguments against stated criteria and produce a reasoned verdict with uncertainty.
Move a scientific object up/down in abstraction or narrow/broaden selected scope dimensions (population, mechanism, context,…
Run structured attack/defense/adjudication over a claim, candidate, criterion set, or current winner. Perspective, target,…
Aggregate criterion or comparison results into an ordered recommendation under an explicit rule.
Abstract relational structure from source domains, map it to the target, validate depth, and instantiate transferable mechanisms.
Direct browser control via CDP for web interaction: automation, scraping, testing, screenshots, and site/app work. Also covers…
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling…