By brian-lai
Implements the PARA (Plan-Review-Execute-Summarize-Archive) methodology for Claude Code, enabling structured task management with persistent context, isolated git worktrees, automated PR creation, and Staff+ engineer code reviews.
Archive the current context to create a clean slate for the next task
Decision helper to determine if PARA workflow should be used for a given request
Execute the active plan by creating an isolated worktree and tracking todos
Display the PARA-Programming quick-reference guide
Initialize PARA-Programming structure in the current project
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.
Structured planning, automated execution, and independent review for AI-assisted development.
PARA turns ad-hoc prompting into a disciplined engineering process: every task gets a plan, every plan gets reviewed, every implementation follows TDD, and every commit maps to a checklist item. The workflow is designed for agent teams — autonomous agents that can research, plan, execute, review, and ship multi-phase projects with minimal human intervention.
# Install the plugin
claude plugin add brian-lai/para-programming-plugin
# Initialize in your project
/para:init
# Run your first task end-to-end
/para:research "add user authentication"
/para:plan "add user authentication"
/para:review --plan
/para:workflow
That's it. /para:workflow handles execution, PR creation, independent review, summarization, and archival — phase by phase — until the work is done.
Research → Plan → Review Plan → Execute → Review PR → Summarize → Archive
For multi-phase work, /para:workflow orchestrates this loop automatically across all phases. It creates worktrees for isolation, commits per checklist item, opens PRs, runs independent subagent reviews, and merges — pausing at phase boundaries for human approval (or running fully autonomously with --auto).
You can run each command individually, or let /para:workflow handle the execution-onward portion automatically:
Manual — run each step yourself:
/para:research → /para:plan → /para:review --plan → /para:execute → /para:review --pr → /para:summarize → /para:archive
Automated — after planning and plan review, hand off to the orchestrator:
/para:workflow runs the full execute → PR → review → summarize → archive → next phase loop. Add --auto for fully autonomous execution, or --skip-review for speed.
| Command | Purpose |
|---|---|
/para:init | Initialize PARA structure in a project |
/para:research <task> | Deep codebase research before planning |
/para:plan <task> | Create a structured implementation plan |
/para:review --plan|--pr | Independent subagent review loop |
/para:execute | Create worktree, extract todos, start TDD execution |
/para:workflow | Orchestrate full cycle across phases |
/para:summarize | Generate post-work summary |
/para:archive | Archive context, clean up worktrees |
/para:status | Show current workflow state |
/para:check | Decision helper: should this task use PARA? |
/para:help | Quick reference guide |
/para:research performs deep codebase exploration and produces a structured document covering architecture, components, API contracts, existing patterns, and gaps. This becomes the primary input for planning — no more ad-hoc exploration during plan creation.
/para:plan creates a collaborative implementation plan through dialogue. It asks clarifying questions, drafts specs and stubs, applies engineering criteria (interface boundaries, graceful degradation, architecture decisions), and runs 2-3 self-review rounds before presenting the plan.
Every implementation step is a checkbox item. Each checkbox becomes one git commit.
/para:review spawns an independent subagent to review plans or PRs. The reviewer categorizes issues as MUST FIX, SHOULD FIX, or NIT, and the review loops — with a fresh subagent each round — until there are no blocking issues. Max 5 rounds with convergence detection.
/para:execute creates an isolated git worktree and works through each checklist item following a TDD cycle: confirm spec → write tests → run red → implement → run green → commit. The checklist text is the commit message.
/para:workflow is the orchestrator. For each phase: execute → create PR → review PR → summarize → merge → archive → next phase. It tracks state in context/context.md for resumability — if interrupted, running it again picks up where it left off.
/para:init)your-project/
├── context/
│ ├── context.md # Active session state
│ ├── data/ # Research docs, specs, datasets
│ ├── plans/ # Implementation plans
│ ├── summaries/ # Post-work reports
│ ├── archives/ # Historical context snapshots
│ └── servers/ # MCP tool wrappers
├── .para-worktrees/ # Git worktree isolation (gitignored)
└── CLAUDE.md # Project-specific context
~/.claude/
└── CLAUDE.md # PARA workflow methodology (shared across all projects)
# Initialize (first time only)
/para:init
npx claudepluginhub brian-lai/para-programming-plugin --plugin paraPARA-Programming methodology - cross-client Agent Skills package for Claude Code, Codex, OpenCode, Cursor, and Gemini CLI
Developer workflow toolkit — JIRA management, Trello boards, and PR creation with ticket references. Invoke with /dev:jira, /dev:trello, or /dev:pr.
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
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.