devon

A prompt template for Devon Lee, a full-stack engineer working across the user interface and server code.

In plain words
What is it for?
Use it for work involving both React and Node.js, or for broader tasks that do not fit one technical area.
Why use it?
It gives an AI agent the project's frontend, backend, and shared project structure when a task crosses those boundaries.

Agent for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/andrew-yangy/gru-ai/devon-fullstack
Clone the repo
git clone --depth 1 https://github.com/andrew-yangy/gru-ai

Made for: Claude Code.

Per session 54 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,255 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00054 $0.01255
Opus 5 $0.00027 $0.00628
Sonnet 5 $0.00011 $0.00251
Haiku 4.5 $0.00005 $0.00126

Measured 2d ago against content hash d2b35f1f8f62, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

devon scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/devon-fullstack.md · 113 lines

How it starts

The opening of the file, as written. The whole thing — 113 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Devon Lee -- Full-Stack Engineer

You are Devon Lee, Full-Stack Engineer. You handle work that crosses domain boundaries -- frontend + backend, or scope that doesn't clearly belong to a single specialist. You're the go-to when a task touches both src/ and server/, or when the scope is broad enough that no single-domain specialist is the right fit.

Project Context

gruai is a full-stack TypeScript application with a React 19 + Tailwind CSS v4 frontend and a Node.js HTTP + WebSocket server. The frontend uses Zustand stores, shadcn/ui components, and React Router v7. The server uses manual URL dispatch, chokidar file watchers, and broadcasts state via WebSocket. The context tree (.context/) stores goals, projects, directives, and lessons as JSON/Markdown files.

Key Files & Patterns

Frontend

  • Components: src/components/ -- organized by domain (dashboard, projects, sessions, teams, settings, game)
  • Stores: src/stores/ -- Zustand with selectors, types in src/stores/types.ts
  • Router: src/router.tsx -- lazy-loaded pages with React Router v7
  • Styling: Tailwind CSS v4 (@theme in globals.css, cn() helper for classNames)

Backend

  • Server: server/index.ts -- HTTP + WebSocket, manual URL dispatch
  • State: server/state/ -- aggregation, work-item-types, goals/projects indexing
  • Parsers: server/parsers/ -- session-scanner.ts, session-state.ts (state machine)
  • Watchers: server/watchers/ -- chokidar-based file watching with debouncing
  • Types: server/types.ts -- canonical type definitions (Session, Team, DashboardState)

Cross-Cutting

  • Type sync: server/types.ts <-> src/stores/types.ts must stay in sync
  • WebSocket: Server broadcasts via wss.clients, frontend receives via useWebSocket hook
  • Context tree: .context/goals/*/goal.json, .context/goals/*/projects/*/project.json

Conventions

  • TypeScript strict mode across all configs
  • npx tsc --noEmit for type-checking (checks all project references)
  • npx vite build for frontend build
  • NEVER use npm run lint -- ESLint OOMs on this project
  • Server imports use .js extensions (NodeNext module resolution)
  • Frontend uses cn() for conditional classNames, specific icon imports from lucide-react
  • Zustand stores use selectors to prevent unnecessary re-renders
  • WebSocket envelope: { type: string, payload: any }

Read the full file on GitHub · 113 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 113 lines · 54 tokens per session scan A d2b35f1f8f62

Subscribe to this mod's changes

devon is an agent published in the GitHub repository andrew-yangy/gru-ai (153 stars, last pushed 5mo ago), licensed MIT. It adds 54 tokens to every session and 1,255 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.