csdlc

Claymore Plugins

Skills and commands for the CSDLC methodology (Claymore Software Development Lifecycle) — a framework for human-AI collaborative software development.

Built for Claude Code. Powered by Foundry as the knowledge layer.

---

Install

# Add the marketplace
/plugin marketplace add danhannah94/claymore-plugins

# Install the plugins
claude plugin install csdlc@danhannah94-claymore-plugins --scope user
claude plugin install foundry@danhannah94-claymore-plugins --scope user

# Activate
/reload-plugins

No clone needed. CC handles caching automatically.

Scopes

Scope	Config file	Use case
user	~/.claude/settings.json	Personal, across all projects
project	.claude/settings.json	Team-shared, committed to git
local	.claude/settings.local.json	Project-specific, gitignored

---

Prerequisites

• Claude Code installed and authenticated
• Foundry instance running (local Docker or deployed) — Foundry is the knowledge layer for CSDLC and is required for full functionality
• GitHub CLI (gh) authenticated — used by /issue and sub-agent PR workflows

---

Commands

Session

Command	What it does
/standup [persona]	Start a CSDLC session. Load context from Foundry, deliver 5-part standup. Optional persona (e.g., /standup clay).
/retro	End session. Write handoff to NEXT.md in Foundry.

Design

Command	What it does
/doc <feature>	Generate a Step 0 design doc. Selects template level (project, epic, subsystem).
/refine <target>	Challenge and refine any artifact — design doc, idea, issue, plan.
/bootstrap [project]	Set up CSDLC for a project — generate WORKFLOW.md, create Foundry space.

Execution

Command	What it does
/breakdown <doc>	Decompose a design doc into implementable stories (Step 1).
/execute <epic>	Work stories with sub-agents, review each PR, wait for approval before merging.
/intern <story>	Hand off a single story to a sub-agent.
/shipit <epic>	Run all stories in an epic autonomously — the Hurricane Pattern.

Quality

Command	What it does
/review <PR>	Structured review checklist — scope, tests, acceptance criteria.
/qa <story>	QA against acceptance criteria with evidence.
/present <scope>	Package completed work for async human review.

Ops

Command	What it does
/issue <title> [--repo owner/repo]	Create a GitHub issue from context.

---

Plugins

CSDLC (csdlc/)

The core methodology plugin. Covers the full CSDLC pipeline:

Refinement (continuous)
    ↓
Step 0: Design Doc        → /doc
Step 1: Story Breakdown    → /breakdown
Step 2: Agent Prompts      → (internal, used by /execute and /intern)
Step 3: Execution          → /execute, /intern, /shipit
Step 4: Review             → /review
Step 5: QA                 → /qa
Step 6: Human Review       → /present

Skills: 13 total — csdlc-standup, csdlc-retro, csdlc-bootstrap, csdlc-doc, csdlc-breakdown, csdlc-execute, csdlc-intern, csdlc-shipit, csdlc-review, csdlc-qa, csdlc-present, csdlc-issue, csdlc-craft-agent-prompt

Bundled templates: Project design, epic design, subsystem design, workflow, agent-base, doc-agent

Foundry (foundry/)

The knowledge layer plugin. Skills for working with Foundry's MCP toolset.

Skills: foundry-mcp-reference (tool patterns + gotchas), foundry-review (annotation conventions for refinement)

---

How It Works

CSDLC is a methodology for human-AI collaborative software development. The key ideas:

1. The process activates on demand. Run /standup to start a CSDLC session. Without it, CC works normally.
2. Foundry is the knowledge layer. Design docs, session state (NEXT.md), stories, and reviews all live in Foundry. This solves the cold start problem — every session starts informed.
3. Skills are Foundry-native. Outputs write to Foundry (annotations, sections, reviews). Inputs read from Foundry.
4. Refinement is the core practice. /refine challenges any artifact before it's acted on. 30 minutes of refinement saves hours of rework.
5. Sub-agents (interns) do the implementation. The AI Lead orchestrates; sub-agents execute in isolated worktrees.

For the full methodology, see the CSDLC Thesis.

---

Project Structure