What is agents-in-the-office CLAUDE.md?

Repository guidance for agents working on a pixel-art office simulation where live AI sessions control characters. It documents the Vue, Rust, and Tauri setup, commands, architecture, and current implementation status.

Use it to understand event flow, subagent handling, frontend and backend structure, and how to install, build, type-check, and lint the project.

It gives coding agents the project context and exact commands they need, reducing guesswork when changing or checking the app.

How gukosowa/agents-in-the-office configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to gukosowa/agents-in-the-office. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/gukosowa/agents-in-the-office/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/gukosowa/agents-in-the-office

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/gukosowa/agents-in-the-office/claude-md/github.svg)](https://agentmods.dev/instructions/gukosowa/agents-in-the-office/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/gukosowa/agents-in-the-office/claude-md"><img src="https://agentmods.dev/badge/instructions/gukosowa/agents-in-the-office/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/gukosowa/agents-in-the-office/claude-md"><img src="https://agentmods.dev/badge/instructions/gukosowa/agents-in-the-office/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 1,366 This file is loaded in full into every session.

When invoked 1,366 The same file — it is already loaded in full.

Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.01366	$0.01366
Opus 5	$0.00683	$0.00683
Sonnet 5	$0.00273	$0.00273
Haiku 4.5	$0.00137	$0.00137

Measured 11d ago against content hash f6f12a25dfc1, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade B, and why

agents-in-the-office CLAUDE.md scanned grade B with 1 finding 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 11d 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- `src/components/AgentSetupDialog.vue` — hook installation dialog (writes hook script + merges into `~/.claude/settings.json`; extracts `agentId` and `agentTranscriptPath` from subagent events)

CLAUDE.md · 70 lines

How it starts

The opening of the file, as written. The whole thing — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.

agents-in-the-office

Pixel-art office simulation where real AI agent sessions control NPC characters.

Stack

Frontend: Vue 3 + TypeScript + Vite 7 + Pinia + TailwindCSS 4
Backend: Tauri 2 (Rust) — file watcher via notify crate, tauri-plugin-fs, tauri-plugin-dialog, tauri-plugin-window-state
Package manager: bun (use bun install, bun run <script> — not npm/npx)
Rust edition: 2024

Dev commands

bun run tauri dev    # start Tauri dev app (Vite + Rust)
bun run build        # frontend-only build (vue-tsc + vite)
bun run tauri build  # release build with installer
vue-tsc -b           # type-check frontend
cargo clippy --all-targets --all-features -- -D warnings  # lint Rust

Architecture

Event pipeline: Claude Code hooks write JSON to ~/.agents-in-the-office/events/ → Tauri Rust watcher picks up files, emits agent-event Tauri events, deletes files → RunView.vue listens → AgentStore routes to driver → driver issues NpcHandle commands → Character executes.

Subagent pipeline: Parent receives subagent_start event → AgentStore.spawnSubagent() derives session ID + transcript path → Rust start_transcript_poll reads JSONL transcript → extracts tool_use/tool_result → emits synthetic events → subagent NPC spawns with badge.

Event persistence: Every event is persisted to IndexedDB (events store). On session reconnect, replayActivityLog() rebuilds the activity panel from stored events. Old events (>24h) are pruned on app mount.

Key source locations:

src/drivers/ — agent driver system:
- types.ts — NpcCommand, AgentEvent, AgentEventType, AgentType, NpcHandle, ActivityMessage, SubagentInfo
- AgentDriver.ts — AgentDriver interface + BaseDriver abstract class (tool→object mapping)
- ClaudeCodeDriver.ts — Claude Code–specific driver (MCP-aware routing, permission wait, subagent handling)
- agentStore.ts — Pinia store for session lifecycle (spawn/despawn/reconnect, stale cleanup, activity log, subagent management). handleEvent() returns Promise<HandleEventResult[]> — one event can yield multiple results (e.g., parent update + subagent spawn)
- npcHandle.ts — createNpcHandle() factory wrapping Character into NpcHandle interface
- activityMessages.ts — formats AgentEvent into ActivityMessage for the activity panel (supports prefix for subagent messages)
- agentBubbleTexts.ts — localized NPC speech/thought bubble texts per tool (en/de, 20+ variants each)
src/classes/Character.ts — NPC entity: states (idle/walking/interacting/conversing/exited), sprite animation, A* pathfinding, command queue, autonomous + external control modes, conversation system, optional badge overlay
src/utils/ — pathfinding.ts (A*), db.ts (IndexedDB — map persistence + event persistence with persistEvent, getSessionEvents, pruneSessionEvents, pruneOldEvents), fileIO.ts (.aito file format), spriteLoader.ts, fill.ts, rmxpAutoTile.ts, vxAutoTile.ts
src/views/RunView.vue — game loop, agent event listener, camera, NPC spawning/despawning, session restore, badge rendering
src/views/EditorView.vue — tile map editor
src/components/AgentActivityPanel.vue — floating panel showing agent session activity log
src/components/AgentSetupDialog.vue — hook installation dialog (writes hook script + merges into ~/.claude/settings.json; extracts agentId and agentTranscriptPath from subagent events)
src/i18n/bubbleTexts.ts — autonomous NPC bubble texts (en/de)
src/stores/mapStore.ts — map state (tile layers, objects, spawn points, collision grid, tileset pool)
src-tauri/src/watcher.rs — file watcher (notify crate, ~/.agents-in-the-office/events/)
src-tauri/src/transcript.rs — JSONL transcript poller for subagents (background thread per subagent, extracts tool_use/tool_result from assistant/user messages, skips Task tool to avoid recursion)
src-tauri/src/lib.rs — Tauri setup, registers set_executable, focus_terminal_window, start_transcript_poll, stop_transcript_poll commands; manages TranscriptPollers state; stops all pollers on window destroy
src-tauri/src/macos_focus.rs — macOS Accessibility API for raising terminal windows

Read the full file on GitHub · 70 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.

11d ago First seen · 70 lines · 1,366 tokens per session scan B f6f12a25dfc1

Subscribe to this mod's changes

agents-in-the-office CLAUDE.md is an instructions file published in the GitHub repository gukosowa/agents-in-the-office (2 stars, last pushed 6mo ago), licensed MIT. It adds 1,366 tokens to every session, about $0.0068 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Game development

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

agents-in-the-office: Instructions file for Claude Code