PMP - GYWD (Get Your Work Done)

Ship faster with AI that remembers your decisions.

The connected context engineering framework for Claude Code — 47 commands · 1,299 tests · 49 lib modules · Zero runtime deps

npx pmp-gywd

Cross-platform: Windows, Mac, Linux

What's New in v5.0?

Connected Intelligence — Web dashboard, semantic memory, multi-model LLMs, CRDT collaboration, enterprise security, and CI/CD integration.

| Feature | Description | |---------|-------------| | Web Dashboard | Next.js 14 planning dashboard with real-time SSE, charts, and analytics | | API Gateway | Express REST + WebSocket API with auth and rate limiting | | Semantic Memory | Zero-dep TF-IDF embeddings for context-aware search and decision similarity | | Multi-Model | Provider-agnostic LLM adapters (OpenAI, Google, Local) with smart routing | | CRDT Collaboration | Multi-user plan editing, decision voting, conflict resolution | | Enterprise | SSO (OIDC/SAML), RBAC, audit logging, SOC2/GDPR compliance | | CI/CD Integration | Pre-merge validation, release notes generation, GitHub/GitLab templates |

          v5.0 Connected Intelligence
┌─────────────────────────────────────────────────┐
│  Web Dashboard    API Gateway    CI/CD Pipeline  │
│  (Next.js 14)    (Express+WS)   (GH Actions)   │
│       │              │              │            │
│       └──────────────┼──────────────┘            │
│                      ▼                           │
│  ┌───────────────────────────────────────────┐  │
│  │         v5.0 Connected Layer              │  │
│  │  Semantic · Models · CRDT · Enterprise    │  │
│  └─────────────────────┬─────────────────────┘  │
│                        │                         │
│  ┌─────────────────────┴─────────────────────┐  │
│  │         v4.0 Agent Runtime                │  │
│  │  Critic · Red Team · Chaos · Skeptic      │  │
│  └─────────────────────┬─────────────────────┘  │
│                        │                         │
│  ┌─────────────────────┴─────────────────────┐  │
│  │         v3.0 Brain + v3.2 Memory          │  │
│  │  Profile · Questioning · GlobalMemory     │  │
│  └─────────────────────┬─────────────────────┘  │
│                        │                         │
│  ┌─────────────────────┴─────────────────────┐  │
│  │         Plugin System (47 commands)        │  │
│  └───────────────────────────────────────────┘  │
└─────────────────────────────────────────────────┘

v4.0: Autonomous Intelligence

Agent Runtime (6 agents), Permission Scanner, Analytics Agents, Self-Grilling, Multi-Agent Coordination, Plugin System, Visual Dashboard

v3.4: Enhanced Experience

MetadataCache, KeywordIndex, graph persistence, VS Code extension, MCP Server

v3.2: Enhanced Learning

GlobalMemory, PatternAggregator, FeedbackCollector, ConfidenceCalibrator, TeamSync

See CHANGELOG.md for full version history. See MIGRATION.md for upgrade guide.

What is GYWD?

The problem: AI generates code that conflicts with your existing decisions, forgets context between sessions, and degrades as projects grow.

The solution: GYWD builds a decision graph of your codebase, maintains persistent context, and now includes autonomous agents that validate, review, and challenge your work.

| Challenge | How GYWD Solves It | |-----------|-------------------| | Lost context between sessions | Decision graph + persistent memory | | AI conflicts with architecture | Checks new code against existing decisions | | "Why was this built this way?" | /gywd:why traces code to decisions | | Unchallenged assumptions | Self-grilling agents question plans | | Security blind spots | Red Team agent probes vulnerabilities |

Getting Started

1. Install

npx pmp-gywd

Choose global (all projects) or local (this project only).

2. Verify

/gywd:help

3. Start Working

| Your Situation | Command | |----------------|---------| | New project from scratch | /gywd:new-project | | Existing codebase | /gywd:bootstrap |

New to GYWD? See the tutorial or workflow examples.

Quick Reference

| Situation | Command | |-----------|---------| | Brand new idea | /gywd:new-project | | Existing codebase | /gywd:bootstrap | | Start of day | /gywd:progress | | Returning after break | /gywd:resume-work | | Why does this code exist? | /gywd:why <file> | | Plan next feature | /gywd:plan-phase | | Execute the plan | /gywd:execute-plan | | Review my approach | /gywd:challenge | | Save state before stopping | /gywd:pause-work |

See /gywd:help for all 47 commands

The Decision Intelligence Paradigm

Understanding WHY

/gywd:why src/utils/result.ts

Output:

DEC-015: Result pattern for error handling [88%]

Rationale: Explicit error handling without exceptions, enabling
type-safe error propagation.

Alternatives Considered:
- Try-catch blocks: Rejected - swallows errors, loses type info
- Error codes: Rejected - easy to ignore, verbose

Trade-offs:
- Gained: Type safety, explicit handling, composability
- Sacrificed: Verbosity, learning curve

Source: Commit a3f2c1 "Implement Result monad for API layer"

Adversarial Review

/gywd:challenge .planning/phases/03-payment/03-01-PLAN.md

Spawns competing agents:

• Critic: Finds logical flaws
• Devil's Advocate: Argues for alternatives
• Red Team: Simulates security attacks
• Chaos Agent: Generates edge cases
• Skeptic: Questions assumptions

Predictive Context

/gywd:anticipate --for "payment integration"

Pre-loads relevant decisions, patterns, and past implementations before you start working.

Library Modules

// Semantic Memory (v5.0)
const { Embedder, SemanticSearch, ContextInjector, DecisionSimilarity } = require('pmp-gywd/lib/semantic');

// Multi-Model LLM (v5.0)
const { OpenAIAdapter, GoogleAdapter, LocalAdapter, ModelRouter } = require('pmp-gywd/lib/models');

// CRDT Collaboration (v5.0)
const { GCounter, PNCounter, LWWRegister, ORSet, PlanEditor, DecisionVoting, ConflictResolver } = require('pmp-gywd/lib/crdt');

// Enterprise (v5.0)
const { SSOManager, RBAC, AuditLog, ComplianceReporter } = require('pmp-gywd/lib/enterprise');

// CI/CD (v5.0)
const { PreMergeValidator, ReleaseNotesGenerator, CIRunner } = require('pmp-gywd/lib/ci');

// Agents (v4.0)
const { CriticAgent, RedTeamAgent, AgentOrchestrator } = require('pmp-gywd/lib/agents');

// Permissions (v4.0)
const { PermissionRouter, RiskScorer } = require('pmp-gywd/lib/permissions');

// Analytics (v4.0)
const { ModelGeneratorAgent, ReviewAgent } = require('pmp-gywd/lib/analytics');

// Memory (v3.2)
const { GlobalMemory, PatternAggregator, TeamSync } = require('pmp-gywd/lib/memory');

// Context & Automation
const { ContextPredictor } = require('pmp-gywd/lib/context');
const { DependencyAnalyzer } = require('pmp-gywd/lib/automation');

49 modules · Zero runtime dependencies · Works offline

Architecture (v5.0)

┌─────────────────────────────────────────────────────────┐
│                    GYWD v5.0 Core                       │
├─────────────────────────────────────────────────────────┤
│                                                         │
│  ┌───────────────────────────────────────────────┐     │
│  │       Web Dashboard · API Gateway · CI/CD      │     │
│  │   (Next.js 14)    (Express+WS)   (GH/GitLab) │     │
│  └─────────────────────┬─────────────────────────┘     │
│                        │                               │
│  ┌─────────┬───────────┼───────────┬─────────┐        │
│  │Semantic │  Multi-   │   CRDT    │Enterpr- │        │
│  │ Memory  │  Model    │  Collab   │  ise    │        │
│  └────┬────┴─────┬─────┴─────┬─────┴────┬────┘        │
│       └──────────┴───────────┴──────────┘              │
│                        │                               │
│  ┌─────────────────────┴─────────────────────────┐    │
│  │           Agent Runtime (v4.0)                 │    │
│  │  Critic · Red Team · Chaos · Skeptic · Devil  │    │
│  └─────────────────────┬─────────────────────────┘    │
│                        │                               │
│  ┌─────────────────────┴─────────────────────────┐    │
│  │  Brain + Memory + Profile + Context             │    │
│  │  GlobalMemory · PatternAggregator · TeamSync   │    │
│  └─────────────────────┬─────────────────────────┘    │
│                        │                               │
│  ┌─────────────────────┴─────────────────────────┐    │
│  │       Plugin System + Command Layer (47)       │    │
│  └────────────────────────────────────────────────┘    │
│                                                         │
└─────────────────────────────────────────────────────────┘

CI/CD Pipeline

• 12 test matrix combinations (3 OS x 4 Node versions)
• 1,299 automated tests with Jest (50 test suites)
• GYWD pre-merge validation — drift detection, decision conflicts, test health
• ESLint with zero external plugins
• Schema/command validation for all 47 commands
• Security scanning with npm audit
• GitHub Actions workflow with PR comments and artifact uploads
• GitLab CI template with reusable job definitions

npm run precommit    # Run all checks locally
npm test             # Run tests
npm run lint         # Run linter
npm run validate:all # Validate schemas and commands

# GYWD-specific CI checks
node lib/ci/ci-runner.js validate          # Pre-merge validation
node lib/ci/ci-runner.js release-notes     # Generate release notes
node lib/ci/ci-runner.js report            # Full CI report

Philosophy

"Code is crystallized decisions. Every function, every pattern exists because someone made a decision with context we've often lost. GYWD makes those decisions explicit and queryable."

v1.x was feature accumulation. v2.0 was unified intelligence. v3.0 is the sophisticated brain - learning, adapting, predicting. v3.4 is enhanced experience - performance, IDE integration, MCP server. v4.0 is autonomous intelligence - agents that challenge, validate, and protect. v5.0 is connected intelligence - web dashboard, multi-model LLMs, collaboration, enterprise, and CI/CD.

The paradigm shift: Instead of generating "plausible code," GYWD generates decision-coherent code that respects the WHY behind your codebase, understands YOU as a developer, actively challenges assumptions, and now connects across tools, models, and teams.

Troubleshooting

Commands not appearing?

• Restart Claude Code to reload commands
• Verify files exist in ~/.claude/commands/gywd/

Commands run but nothing happens?

• Check .planning/ directory exists
• Run /gywd:init if missing

Plan execution stuck?

• Run /gywd:progress to see current state
• Check .planning/STATE.md for position

Lost context mid-session?

• Run /gywd:resume-work to restore
• Or /gywd:progress to see what's next

Need latest version?

npx pmp-gywd@latest

Contributing

Issues and PRs welcome at github.com/cyberbloke9/pmp-gywd.

Before submitting:

• Run npm test (1,299 tests must pass)
• Run npm run lint

See docs/CONTRIBUTING.md for full guidelines.

License

MIT License - See LICENSE

Understand decisions. Challenge assumptions. Connect intelligence. Ship coherent code.

Built with connected intelligence by cyberbloke9