By gerquinn1978
The Grounded Vibe Methodology (GVM): a citation-based software development lifecycle that anchors AI output to named experts and their published works. Pipeline skills: /gvm-init, /gvm-site-survey, /gvm-impact-map, /gvm-requirements, /gvm-test-cases, /gvm-tech-spec, /gvm-design-review, /gvm-walking-skeleton, /gvm-build, /gvm-code-review, /gvm-test, /gvm-explore-test, /gvm-doc-write, /gvm-doc-review, /gvm-deploy. Standalone and shared: /gvm-analysis (privacy-preserving exploratory data analysis), /gvm-experts, /gvm-status, /gvm-design-system. Includes curated expert reference files, activation tracking, expert scoring, calibrated review with dual-review correction, outside-in TDD discipline, mock-budget enforcement, wiring-matrix audit, chunk-acceptance smoke gate, and a methodology changelog with defect-class citations.
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Privacy-preserving exploratory data analysis. Claude orchestrates a Python engine that reads the user's data, produces a grounded findings report with headline insights, provenance, and a methodology appendix. Raw rows never enter Claude's context.
Use when reviewing implementation code for quality, security, and spec adherence. Triggered by /gvm-code-review command, requests to review code, or after completing a build phase. Dispatches parallel defect-class panels as subagents, grounded in the project's spec-declared specialists.
Use when preparing a release — version bump, release notes, documentation accuracy check, changelog, and git tag. Triggered by /gvm-deploy command, requests to prepare a release, cut a version, or generate release notes. Does NOT execute deployment — that is organisation-specific.
Use when reviewing whether a design (data model, UI/UX, API) correctly fulfils requirements. Triggered by /gvm-design-review command, requests to review a database schema against requirements, check if a UI covers all user journeys, or verify API contracts. Also triggered by /gvm-site-survey when it identifies design concerns. Dispatches parallel defect-class panels as subagents, grounded in the project's spec-declared specialists.
Shared design references for all GVM skills. Not invoked directly by users — other skills load its reference files at runtime. Contains the Tufte/Few HTML design spec, expert reference files, shared rules, pipeline contracts, and scoring methodology. For expert scoring and roster management, use /gvm-experts instead.
Citation-based expert grounding for AI-assisted software development. Name the expert, cite the work, get focused output instead of undifferentiated training data.
grounded-vibe-methodology/ — a Claude Code plugin that implements GVM as a phased pipeline. Install it and every pipeline phase activates named authorities for requirements, test design, architecture, build, and review.
See grounded-vibe-methodology/README.md for installation and usage.
All whitepapers are in whitepapers/. Open in a browser. Suggested reading order:
| # | Paper | What it covers |
|---|---|---|
| 1 | The Grounded Vibe Methodology | The full thesis: citation-based grounding, expert model, pipeline, experiment, attribution argument |
| 2 | Well-Engineered Software Needs an End-to-End Pipeline | The structural decisions and durable artefacts that distinguish the GVM pipeline, independent of expert grounding |
| 3 | Grounding Produces Measurably Better Output | The controlled experiment: 180 dashboards, three models (Claude, GPT, Gemini), 1,080 blind reviews, dose-response results |
| 4 | Why Citation-Based Activation Works | Transformer-level mechanism: why naming an expert changes what the model produces |
| 5 | Verifying AI-Generated Code | Five-layer verification architecture: property-based testing, mutation testing, security testing, stack-aware tooling |
| 6 | Beyond the Pipeline | Where the methodology goes once the pipeline is working: standalone tools, document types beyond pipeline artefacts, Tier 2b cross-project compounding, the reflexive case study, and worked sketches in non-software disciplines |
Each paper includes a "This paper in context" guide linking to the others.
180 market risk dashboards across three models (Claude, GPT, Gemini), three grounding conditions, 1,080 blind reviews. The experiment paper is in whitepapers/. The experiment configs (exact prompts, grounding conditions, review rubrics) are in experiments/. The results CSVs (scores and group mappings for all three models under both review conditions) are in experiments/results/.
case-studies/case-study-assay.html — a full end-to-end case study of building Assay (the experiment runner) with GVM. Covers requirements through release, with expert patterns traceable in the production code.
Eight training documents in training/:
| Document | What it covers |
|---|---|
| Getting Started | Install, pipeline walkthrough, quick reference |
| Expert Grounding Workshop | Why citations work, practice exercises, where grounding is effective |
| Review Calibration Training | Progressive calibration, dual review, build check tiers, practice exercises |
| Triage Workshop | Practising the Finding Quality Gate across document, code, and design reviews; fix/defer/dismiss decisions with worked examples |
| Writing Requirements That Pass | Acceptance-criteria-first requirements that survive /gvm-test-cases Phase 4; five common failure modes with before/after examples |
| Standalone /gvm-doc-write | Producing presentations, strategy, newsletters, and prose grounded in named experts — outside the pipeline |
| Engineering Leader's Guide to GVM | Adopting the methodology — what it changes, what it costs, how to roll it out |
| /gvm-analysis Training | Privacy-preserving exploratory data analysis with Tukey, Cleveland, Hyndman & Athanasopoulos grounding |
blog-announcing-gvm.html — a short introduction to the methodology for engineers who want the argument before the detail.
Install the plugin:
git clone <repository-url>
claude plugin add grounded_vibe_methodology/grounded-vibe-methodology
Verify it works:
/gvm-status
Start a project:
/gvm-requirements
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimnpx claudepluginhub gerquinn1978/gvm --plugin grounded-vibe-methodologyComprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Intelligent draw.io diagramming plugin with AI-powered diagram generation, multi-platform embedding (GitHub, Confluence, Azure DevOps, Notion, Teams, Harness), conditional formatting, live data binding, and MCP server integration for programmatic diagram creation and management.
Comprehensive PR review agents specializing in comments, tests, error handling, type design, code quality, and code simplification