By drbothen
Verified Spec-Driven Development (VSDD) dark factory for software — full SDLC pipeline: brownfield ingest, spec crystallization, story decomposition, TDD delivery, adversarial review, holdout evaluation, formal verification, and release gating.
Use when auditing specs, designs, or implementations for WCAG AA/AAA accessibility compliance at any pipeline checkpoint.
Fresh-context adversarial reviewer for specs and implementation. Finds gaps, contradictions, missing edge cases, and unstated assumptions. Uses different model for genuine perspective diversity. Cannot see prior review passes.
Use when designing system architecture from domain specs and PRDs, producing ADRs, subsystem decomposition, and verification-ready design artifacts.
Use when synthesizing the L2 Domain Specification from a product brief and domain research, producing sharded capability, entity, and invariant docs.
Use when performing constructive code review on a PR or diff with cognitive diversity from a different model family, producing classified findings without modifying code.
Opt in to the VSDD factory persona for this project. Detects the host platform, persists the detected platform string, and writes `.claude/settings.local.json` to set the orchestrator as the default main-thread agent. Reversible via `/vsdd-factory:deactivate`.
Launch a fresh-context adversarial review of specs or implementation. Uses the adversary agent with information asymmetry to find gaps, contradictions, and missing edge cases. Minimum 3 clean passes to convergence.
Reviews AGENTS.md files for compliance with Dark Factory agent design principles. Checks token budget, contradictions, negative examples, FACTORY.md duplication, tool profile mismatches, and structural compliance with the canonical template.
Reads product analytics data (if available) to identify feature adoption, error patterns, and usage signals. Optional -- only runs if analytics sources are configured. Does not implement telemetry in the product.
Scans the project for existing planning artifacts (brief, PRD, architecture, UX spec, stories), validates their quality, identifies gaps, and routes to the correct pipeline entry point. This is the universal front-end for the VSDD pipeline — it replaces the assumption that the human always arrives with a finished product brief.
Uses power tools
Uses Bash, Write, or Edit tools
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 claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Verified Spec-Driven Development (VSDD) -- a dark factory for software, packaged as a Claude Code plugin.
Verified Spec-Driven Development is a software engineering methodology that fuses three proven paradigms into a single AI-orchestrated pipeline:
The methodology produces software with full traceability from product brief through domain model, behavioral contracts, acceptance criteria, test cases, implementation, and formal proofs. Every line of code can answer "why does this exist?" by tracing back to a spec requirement.
The VSDD contract chain traces every artifact through four specification levels:
L1 Product Brief
-> L2 Domain Spec Capability (CAP-NNN)
-> L3 Behavioral Contract (BC-S.SS.NNN)
-> Story Acceptance Criterion (AC-NNN)
-> Test Case
-> Implementation
-> L4 Verification Property (VP-NNN)
-> Formal Proof
Quality is not declared by the builder -- it is measured by five independent convergence dimensions (spec, tests, implementation, verification, holdout), each gated by quantitative thresholds. The pipeline terminates when further adversarial review produces only hallucinated findings.
The vsdd-factory plugin installs the complete VSDD pipeline into any Claude Code session:
/plugin marketplace add drbothen/claude-mp
/plugin install vsdd-factory@claude-mp
Migrating from a pre-rc.7 install? The marketplace was at
drbothen/vsdd-factorythrough rc.6 but suffered from a silent install bug (seedocs/guide/plugin-marketplace-architecture.mdfor the full story). Run/plugin marketplace remove vsdd-factoryfirst, then adddrbothen/claude-mpand install via the new marketplace as shown above.
/plugin marketplace update claude-mp
/plugin update vsdd-factory@claude-mp
/vsdd-factory:factory-health # verify .factory/ worktree is mounted
/vsdd-factory:setup-env # verify toolchain (first time or after changes)
Then read .factory/STATE.md to understand where the pipeline left off.
claude --plugin-dir ./plugins/vsdd-factory
graph TD
P0["Phase 0\nBrownfield Ingest"]
P1["Phase 1\nSpec Crystallization"]
P2["Phase 2\nStory Decomposition"]
P3["Phase 3\nTDD Implementation"]
P4["Phase 4\nHoldout Evaluation"]
P5["Phase 5\nAdversarial Refinement"]
P6["Phase 6\nFormal Hardening"]
P7["Phase 7\nConvergence & Release"]
P0 -->|"existing codebase"| P1
START(("New project")) -->|"greenfield"| P1
P1 -->|"specs locked"| P2
P2 -->|"stories ready"| P3
P3 -->|"wave merged"| P4
P4 -->|"holdout pass"| P5
P5 -->|"findings fixed"| P3
P5 -->|"converged"| P6
P6 -->|"proofs pass"| P7
P6 -->|"issues found"| P5
P7 -->|"not converged"| P3
P7 -->|"CONVERGED"| RELEASE(("Release"))
npx claudepluginhub drbothen/claude-mp --plugin vsdd-factoryICS/OT Security Operations — CVE enrichment, event investigation, MITRE ATT&CK mapping, and adversarial quality review with convergence-driven analysis.
Comprehensive 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.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.