weside-ai

we — Agentic Product Ownership for Claude Code

The first Agentic Product Ownership toolkit for Claude Code. Covers the full product development chain: from story refinement through development, code review, and CI automation.

What is Agentic Product Ownership?

Unlike AI coding assistants that help developers write code, Agentic Product Ownership focuses on the strategic side: shaping products, not just building them. From vision alignment through story creation to delivery tracking.

Learn more at agenticproductownership.com

Install

/plugin marketplace add weside-ai/claude-code-plugin
/plugin install we@weside-ai

Quick Start

# Set up your project (auto-detects stack + ticketing tool)
/we:setup

# Create a story with implementation plan
/we:refine "Add user authentication"

# Implement it end-to-end (autonomous pipeline)
/we:story PROJ-1

---

The Development Pipeline

Three phases, clear responsibilities:

Idea → /we:refine → /we:story → User merges → Done
         (manual)    (autonomous)   (manual)

Phase	Who	What	Command
Planning	User + Claude (interactive)	Story + Plan	/we:refine
Development	Claude (autonomous)	Code → Review → Test → Docs → PR → CI	/we:story
Delivery	User (manual)	Review PR, merge, close ticket	GitHub / Ticketing

Phase 1: Planning

/we:refine is interactive. Claude asks questions, you make decisions. The output is:

• Ticket (minimal): "As X I want Y so that Z" + link to plan
• Plan (detailed): docs/plans/{TICKET}-plan.md with acceptance criteria, phases, tests, security review

Phase 2: Development

/we:story runs autonomously — you sit back and watch:

/we:story {TICKET}
  ├── Load story + plan, create worktree
  ├── /we:develop (implement phase by phase)
  ├── AC Verification (every criterion checked with evidence)
  ├── Simplify (requires code-simplifier plugin)
  ├── Quality Gates in PARALLEL:
  │     /we:review + /we:static + /we:test + CodeRabbit
  ├── /we:docs (auto-update documentation)
  ├── /we:pr (prerequisite check: all gates passed?)
  ├── /we:ci-review (collect → fix → push, max 3 cycles)
  └── Ticket → "In Review"

Every step writes a checkpoint to SQLite. If interrupted, /we:story resumes from where it left off.

Phase 3: Delivery

You get a PR with all ACs implemented, tests passing, code reviewed, docs updated, CI green. You review, merge, close the ticket. Claude never merges PRs or closes tickets.

Robustness

Feature	What it does
Checkpoints	Resume after interruption (SQLite)
Circuit Breaker	3 failures in same phase → stop and ask
Batch-Fix	Collect ALL findings, fix in one commit, push once
Reality Check	Warn if plan is stale vs. recent code changes

---

All Skills

Pipeline Skills (called by /we:story)

Skill	What it does
/we:develop	Implement code from plan, phase by phase
/we:review	Code review (runs as background agent)
/we:static	Lint, format, type check (auto-detects stack)
/we:test	Run tests with coverage (auto-detects framework)
/we:docs	Auto-detect and update affected documentation
/we:pr	Create PR (validates all quality gates passed)
/we:ci-review	Collect CI/review findings, batch-fix, push

Standalone Skills

Skill	What it does
/we:setup	Project onboarding (3 questions, auto-detection)
/we:refine	Create stories with acceptance criteria and plans
/we:arch	Architecture guidance, ADR creation
/we:sm	Process optimization, retrospectives
/we:doc-review	Documentation structure and quality review
/we:doc-check	Documentation content consistency check
/we:find-dead-code	Remove dead code from Python backends
/we:materialize	Load weside Companion identity (optional)

Background Agents (called by skills)

Agent	Purpose
code-reviewer	Diff-based code review, AC alignment, max 10 issues
static-analyzer	Lint, format, types — auto-detects your stack
test-runner	Tests with coverage gates — auto-detects framework
pr-creator	PR with prerequisite checkpoint validation
doc-manager	Auto-detect and update project documentation

---

Stack Detection

Skills auto-detect the project stack:

File	Stack	Lint	Types	Tests
pyproject.toml	Python	ruff	mypy	pytest
package.json	Node.js	eslint	tsc	jest/vitest
Cargo.toml	Rust	clippy	(built-in)	cargo test
go.mod	Go	golangci-lint	(built-in)	go test

Ticketing

Skills detect the available ticketing tool automatically: