Stats
Actions
Available In
Tags
More by nwleedev
system
0·
Claude Code skills and agent workflows for development documentation.
1w
nwleedevBy nwleedev
Claude Code Harness System — plan enforcement, quality gates, domain knowledge, and session continuity
An interactive professor agent that systematically teaches a specific domain from the ground up. When users lack domain knowledge, it conducts lectures based on the deep-study skill and records learning materials in .claude/feeds/.
A specialized research agent for domain investigation and harness skill generation/verification. Used when the harness-engine skill delegates to a sub-agent.
An agent that performs deep analysis of plan execution readiness. Evaluates ambiguity, missing information, and whether additional questions are needed.
An agent that performs project-level research. Provides evidence-based answers for general research questions such as technology selection, library comparison, architecture decisions, and best practice verification.
An agent that analyzes the transcript at turn end to generate a narrative context summary.
Use when the user wants to learn a domain from scratch, needs structured teaching, or says they have no knowledge about a topic. Provides step-by-step curriculum, assessments, and progressive learning.
Response rules for failures — errors, test failures, build failures, type errors, runtime exceptions, blocked states, and problem resolution. Use when encountering errors, test failures, or blocked states.
A skill that creates or enhances `.claude/skills/harness-<domain>-<name>.md` harness skills tailored to the user's work domain. The main agent handles decisions and user interaction, while sub-agents perform domain research and harness generation.
Research methodology for questions requiring investigation — technology selection, library comparison, architecture decisions, best practice verification. Used by both the project-researcher agent and inline research.
Socratic thinking principles. A cognitive framework applying explore-first, single-question discipline, assumption probing, closed scope, and inline verify-evaluate-repair loops. Referenced by plan-readiness-checker; also user-invocable.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
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.
A configuration/automation framework that installs automatic rules, quality gates, and domain knowledge on Claude Code.
Not a code library. It's a configuration layer that sits on top of Claude Code.
| Claude Code Default | With Harness |
|---|---|
| Starts coding/writing immediately on request | Enforces planning before editing (hooks block) |
| Repeats library anti-patterns | Auto-suggests domain rules when files are read |
| Large changes pass without review | 2-4 single review, 5+ parallel quality review |
| Context lost after long conversations | Auto-saves/restores context on session end/compaction |
| Plan quality is inconsistent | Auto-validates completeness when plan finishes |
| Risky git commands possible | Blocks destructive commands like --force, reset --hard |
Applies to all work using Claude Code — not just coding, but research, marketing, design, learning, and more.
| Audience | Example Projects | Details |
|---|---|---|
| Frontend Developers | React/Next.js web apps, UI component libraries | Quick Start |
| Backend Developers | FastAPI REST API, Go microservices | Quick Start |
| Researchers | Market research, competitor analysis, tech trends | Quick Start |
| Marketers | Campaign planning, A/B test design, ad copy | Quick Start |
| Architects | System architecture docs, RFCs, ADRs | Quick Start |
| Learners | Kubernetes intro, React learning, ML basics | Quick Start |
Install as a Claude Code plugin without copying files to your project.
claude plugin install nwleedev/engine
For per-project settings, create .claude/engine.config (optional):
# .claude/engine.config
REVIEW_AGENTS="domain,structure" # Review perspectives (comma-separated)
RESEARCH_PERSPECTIVES="pro,con" # Research perspectives (comma-separated)
Copies all files to your project's
.claude/directory.
# Install harness to project (one-liner, no git needed)
sh -c "$(curl -fsSL https://raw.githubusercontent.com/nwleedev/engine/main/install.sh)" -- ~/my-project
# Run Claude Code
cd ~/my-project && claude
Update: bash .claude/scripts/update.sh or re-run the same command.
git clone https://github.com/nwleedev/engine.git /tmp/engine
/tmp/engine/.claude/scripts/bootstrap.sh --source /tmp/engine --target ~/my-project
rm -rf /tmp/engine
cd ~/my-project && claude
| Item | Plugin | Standalone |
|---|---|---|
| Install | claude plugin install | curl one-liner |
| Update | Automatic (plugin manager) | update.sh manual |
| Multiple projects | One install, shared | Install per project |
| Customization | .claude/engine.config | .claude/settings.local.json |
| Skill prefix | /engine:deep-study | /deep-study |
Prerequisites: Claude Code CLI. Standalone install additionally requires curl, tar, jq. See Getting Started for detailed setup.
| Hook | Trigger | Action |
|---|---|---|
| Plan enforcement | Before file edit | Blocks editing without a plan |
| Plan quality validation | On plan mode exit | Auto-checks file existence, required sections, ambiguity |
| Harness suggestion | After file read | Detects domain keywords in file content → suggests related rules |
| Edit tracking | After file edit | Tracks modified file count; 2-4 single / 5+ parallel review |
| Session snapshot | On session end | Saves work state to .claude/sessions/ |
| Context restore | On session start/compaction | Restores context from previous snapshots |
| Skill | Invocation | Purpose |
|---|---|---|
| harness-engine | /harness-engine | Auto-generate project-specific domain rules |
| deep-study | /deep-study | Step-by-step learning protocol (assess→design→lecture→verify) |
| failure-response | /failure-response | Error response methodology (root cause first) |
| research-methodology | /research-methodology | Structured research framework |
| socratic-thinking | /socratic-thinking | Critical thinking principles (explore first, verify assumptions) |
npx claudepluginhub nwleedev/engineClaude Code skills and agent workflows for development documentation.
nwleedevHarness for Claude Code — skills, /harness:* slash commands, persona subagents, lifecycle hooks, and MCP tools without per-repo `harness setup`. Sibling plugins exist for Cursor, Gemini CLI, and Codex.
Intense-VisionsComplete collection of battle-tested Claude Code configs from an Anthropic hackathon winner - agents, skills, hooks, and rules evolved over 10+ months of intensive daily use
marshall0524Verification-first engineering toolkit for Claude Code. 15 skills across a 5-phase spine (Investigate → Design → Implement → Verify → Ship), 8 specialist agents, an interactive setup wizard. Every skill has rationalizations + evidence requirements. Built for senior ICs and tech leads.
duthahoSession harness plugin for Claude Code workflow automation
team-attentionHarness engineering for Claude Code — hook-enforced dual review, state-machine gates, and fail-closed safety where it counts.
sd0xdevMulti-agent /workflow development pipeline (planner → plan-review → coder → code-review) with typed handoff contracts, lifecycle hooks, and MCP servers.
hex0xdeadbeef