Disciplined Process

Disciplined Process Plugin

A rigorous, traceable AI-assisted development workflow for Claude Code. Inspired by the Rue language development process.

Version: 2.3.0

See the project README for overview and installation.

The Workflow

A 7-phase loop: Orient → Specify → Decide → Test → Implement → Review → Close

Phase	Action	Command
1. Orient	Find and claim work	/dp:task ready
2. Specify	Write requirements with [SPEC-XX.YY] IDs	/dp:spec add
3. Decide	Record architectural choices	/dp:adr create
4. Test	Write failing tests with @trace markers	—
5. Implement	Make tests pass, add @trace to code	—
6. Review	Self-review + adversarial review	/dp:review
7. Close	Complete task, commit	/dp:task close

See references/workflow.md for detailed phase descriptions.

Commands

Command	Description
/dp:init	Initialize project with wizard
/dp:task	Task tracking (ready, create, show, update, close, discover)
/dp:spec	Specification management (create, add, link, coverage)
/dp:adr	Architecture Decision Records (create, list, status, link)
/dp:review	Run code review (with optional adversarial mode)
/dp:session	Session management (Chainlink only)
/dp:migrate	Migrate between task trackers
/dp:trace	Traceability validation
/dp:help	Show help and workflow reference

Task Commands

/dp:task ready                          # Show available work
/dp:task create "Fix bug" -t bug        # Create a task
/dp:task show <id>                      # View task details
/dp:task update <id> --status in_progress
/dp:task close <id> --reason "Fixed"
/dp:task discover "Edge case" --from <id>  # File discovered work
/dp:task tree                           # Dependency visualization (Chainlink)
/dp:task blocked                        # Show blocked issues

Session Commands (Chainlink only)

/dp:session start "Working on auth"     # Start tracked session
/dp:session work <id>                   # Claim task and start timer
/dp:session status                      # Show current session
/dp:session end --notes "Handoff notes" # End session with context

Migration Commands

/dp:migrate beads-to-chainlink          # Migrate from Beads to Chainlink
/dp:migrate chainlink-to-beads          # Migrate from Chainlink to Beads
/dp:migrate --dry-run                   # Preview without changes

Traceability Commands

/dp:trace coverage                      # Show spec-to-code coverage
/dp:trace validate                      # Verify all traces resolve
/dp:trace find SPEC-01.03               # Find code implementing spec

Spec Commands

/dp:spec create 02 "Authentication"     # Create spec section
/dp:spec add 02 "Users MUST authenticate before access"
/dp:spec coverage                       # Check trace coverage
/dp:spec list                           # List all specs

ADR Commands

/dp:adr create "Use JWT for authentication"
/dp:adr list                            # List all ADRs
/dp:adr status 0001 accepted            # Update ADR status
/dp:adr link 0001 SPEC-05.12            # Link ADR to spec/task

Configuration

Enforcement Levels

Level	Behavior
Strict	Hooks block commits without tests/traces
Guided	Hooks warn but don't block
Minimal	Skills on demand, no enforcement

See references/enforcement-config.md for per-hook overrides and team recommendations.

Task Tracking Providers

Provider	Requirements	Key Features	Availability
Beads (recommended)	Git, bd CLI	Git-native, dependencies, MCP server	Public
Builtin	Claude Code v2.1.16+	Zero setup, dependency tracking	Built-in
Chainlink	chainlink CLI	Sessions, time tracking, tree view	Private
GitHub Issues	gh CLI	Team collaboration, web UI	Public
Linear	linear CLI	Native priorities, projects	Public
Markdown	None	No dependencies, human-readable	Built-in
None	None	External tracking	-

See commands/task.md for the full feature matrix and provider details.

Adversarial Review

Enable VDD-style adversarial code review with a fresh-context model:

/dp:review --adversarial               # Run adversarial review
/dp:review --adversarial --max-iterations 3

The adversarial reviewer (Sarcasmotron) uses Gemini with fresh context to:

• Detect hallucinations (invalid line numbers, missing functions)
• Find logic errors and edge cases
• Challenge assumptions
• Iterate until issues are resolved or accepted

See commands/review.md for configuration options.

Project Structure After Init

project/
├── .claude/
│   ├── dp-config.yaml      # Plugin configuration
│   ├── settings.json       # Hooks configuration
│   ├── rules/              # Language-specific rules
│   │   ├── python.md
│   │   └── typescript.md
│   └── traceability/       # Auto-generated trace index
│       └── index.json
├── docs/
│   ├── spec/               # Specifications
│   │   └── 00-overview.md
│   └── adr/                # Architecture Decision Records
│       ├── template.md
│       └── 0001-adopt-disciplined-process.md
├── CLAUDE.md               # Project context for Claude
└── .beads/                 # Task tracker (if using Beads)

Configuration File

Full dp-config.yaml example:

version: "2.0"

project:
  name: "my-project"
  languages:
    - typescript
    - python

# Issue tracker selection
task_tracker: beads  # beads | builtin | chainlink | github | linear | markdown | none

# Beads-specific configuration (when task_tracker: beads)
beads:
  auto_sync: true
  daemon: true

# Enforcement level
enforcement: guided  # strict | guided | minimal

# Adversarial review configuration
adversarial_review:
  enabled: true
  model: gemini-2.5-flash
  max_iterations: 5

# Spec configuration
specs:
  directory: docs/spec/
  id_format: "SPEC-{section:02d}.{item:02d}"

# ADR configuration
adrs:
  directory: docs/adr/
  template: .claude/templates/adr.md

# Degradation behavior
degradation:
  on_tracker_unavailable: warn  # warn | skip | fail
  on_hook_failure: warn

Skills (Auto-Invoked)

These skills activate automatically based on context:

• disciplined-workflow: Core process orchestration
• spec-tracing: Specification format and traceability
• tdd-methodology: Test-first patterns for all test types
• adr-authoring: Decision record format

Troubleshooting

See the project README for installation and update issues.

/dp:init fails

• Ensure you're in a valid project directory
• Check write permissions
• Use /dp:init --force to overwrite existing files

Trace markers not found

Trace markers must follow @trace SPEC-XX.YY format:

// @trace SPEC-01.03
function validateInput(data: string): boolean {
  // ...
}

Provider CLI issues

If task commands fail, check your provider's CLI:

• Chainlink: chainlink --version (requires .chainlink/ initialized)
• Beads: bd --version (requires git repo with .beads/)
• GitHub: gh auth status
• Linear: linear --version

Adversarial review not working

• Ensure adversarial_review.enabled: true in dp-config.yaml
• Check that Gemini API access is configured in rlm-claude-code
• Use /dp:status to check component health

Migration issues

• Use --dry-run first to preview changes
• Check .claude/dp-migration-map.json for ID mappings after migration
• Spec references are updated automatically

More Information

• Commands Reference
• Skills Reference
• Workflow Reference
• Enforcement Config
• Adversarial Review
• Migration Guide