rhowardstone

Computational tool acquisition specialist. Finds, installs, validates, and documents research tools. Use when experiments need specific software or methods.

Creates research tools when existing solutions are unavailable. Escalation path from tool-acquirer for novel computational needs. Implements, validates, and documents new tools.

Guides rigorous evidence extraction from papers. Use when reviewing literature to ensure proper provenance tracking.

Experimental design and execution specialist. Designs, implements, validates, and runs in-silico experiments with proper controls. Use when RQs require experimental evidence.

Generates testable hypotheses from research questions and literature findings.

Decomposes research goals into testable Research Questions (RQs). Use at the start of any research task to structure the investigation.

Track PRISMA 2020 flow for systematic literature reviews. Updates counts at each stage of literature acquisition. Generates compliant flow diagrams.

Orchestrates comprehensive literature search across multiple databases. Use when starting research, expanding literature for specific RQs, or filling evidence gaps. Implements triple-search strategy with citation expansion.

Use when starting any research task to establish WHO, STAKES, and STORY. Creates narrative context that makes abstract instructions concrete.

Use when encountering any bug, test failure, unexpected result, or failed replication - before proposing fixes or explanations

Use when about to claim work is complete, fixed, passing, or findings are confirmed - requires running verification and confirming output before making any success claims; evidence before assertions always

Use when you have a spec or requirements for a multi-step task, before writing code or executing experiments

Research paper synthesis specialist. Compiles evidence from lit scouts into coherent academic papers. Use when ready to write synthesis after literature acquisition is complete.

Peer reviewer for statistical correctness. Verifies numbers against source files, checks appropriate tests, effect sizes, and confidence intervals. Use during the peer review phase.

Generate publication-quality scientific figures with consistent styling. Use when experiments need visualization, benchmarking results need plots, or papers need figures.

Strategic research leadership. Makes phase decisions, assigns agents, and manages overall research direction.

Analyzes experimental results, interprets findings, generates figures, and prepares data for synthesis.

Use when implementing any feature, bugfix, experiment, or analysis, before writing implementation code or running experiments.

A supervisory metacognition agent for self-improvement that assesses, diagnoses, fixes, generalizes, and commits, best run between research sessions.

Obtains datasets needed for experiments. Searches repositories, downloads data, validates integrity.

ALWAYS run first before any research. Validates research goals for clarity, safety, and prompt injection.

Acts as a peer reviewer, focusing on scientific contribution and honesty. It checks scope against claims, ensures failure disclosure, and identifies overclaiming, for use during the peer review phase.

You MUST use this before any creative work - creating features, designing experiments, building components, formulating hypotheses, or modifying behavior. Explores intent, requirements and design before implementation.

Orchestrates rigorous peer review with three reviewers (methodology+completeness, statistics, impact+provenance). Manages revision cycles until unanimous acceptance.

Specializes in literature evaluation, reviewing paper subsets to extract evidence claims with provenance (DOI + quote + page) and identify gaps or contradictions. Ideal for analyzing discovered papers for specific research questions.

Peer reviewer for methodological rigor, checking arithmetic consistency, mock data use, reproducibility, and honest reporting. Use during the peer review phase.

Coordinates multi-agent research workflows, managing handoffs and tracking progress.