Paste this into Claude Code:
Update this skill: https://github.com/Solnest-AI/solnest-coding-agent
Claude pulls the latest version and installs it. This plugin stores no settings of its own, so there's nothing of yours to lose — just quit and reopen Claude Code afterwards so it reloads.
Claude, if you are reading this page because the user asked you to update this skill: fetch and follow
UPDATE.mdin this repo, top to bottom.
A 10-agent coding plugin for Claude Code — a master orchestrator plus 9 engineering specialists, with automated quality gates baked in.
This is the same agent system Solnest AI uses every day to build client projects.
Inside Claude Code, type:
/plugin marketplace add Solnest-AI/solnest-coding-agent
/plugin install solnest-coding-agent@solnest
Restart Claude Code, then type /agents — you should see coding-agent plus 9 specialists.
Easiest path: download INSTALL-SOLNEST-CODING-AGENT.md from this repo, drag it into Claude Code, and say "set this up" — Claude walks you through everything, checks your prerequisites, and verifies the install.
| Agent | What it does |
|---|---|
| coding-agent | Master orchestrator — runs THE LOOP: clarify → plan + Definition of Done → build → test → verify → repeat until done |
| ai-engineer | Claude API, prompts, RAG, embeddings, agent design |
| backend-architect | Python, Node.js, APIs, databases, cloud architecture |
| frontend-developer | React, Next.js, Tailwind, mobile |
| ui-designer | Layouts, design systems, accessibility, UX |
| code-reviewer | Code review, debugging, refactoring, build errors |
| devops | CI/CD, Docker, Terraform, Kubernetes, deployment |
| security-auditor | Security review, threat modeling, dependency audits |
| performance-engineer | Profiling, benchmarking, optimization, caching |
| project-shipper | Launch checklists, risk assessment, client delivery |
Plus two automated quality gates (no setup needed):
- Secret scan — blocks any
git committhat contains hardcoded API keys or credentials. - Auto-test loop — runs your project's tests after every code edit and feeds failures straight back to Claude to fix.
The coding-agent doesn't just write code — it runs a closed loop: it asks clarifying questions up front (only when getting it wrong would be expensive), writes a Definition of Done, builds, tests after every edit, then dispatches fresh-context critics (code review + security audit, scaled up for complex work) that never review their own code, and loops until the Definition of Done is met.
See solnest-coding-agent/README.md for full agent docs.
/plugin update solnest-coding-agent
New versions are announced in the Solnest AI community — see CHANGELOG.md.
- Claude Code (latest)
gitandjq(the installer checks these for you)
Built by Solnest AI — we help business owners implement AI so they can work on their business, not in it.
MIT — see LICENSE. Copyright (c) 2026 Solnest AI.