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.
npx agentmods add instructions/webteractive/zetty/agents-mdgit clone --depth 1 https://github.com/webteractive/zettyWhat 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 | $0.22635 | $0.22635 |
| Opus 5 | $0.11317 | $0.11317 |
| Sonnet 5 | $0.04527 | $0.04527 |
| Haiku 4.5 | $0.02263 | $0.02263 |
Grade D, and why
zetty AGENTS.md scanned grade D with 3 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- **Claude** — additive hooks in `~/.claude/settings.json` (UserPromptSubmit→running, Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
one mechanism Zetty owns is the swap helper, which used to `rm -rf` the target Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
A path can come from arbitrary terminal output (a log, a `curl` response), so Copies of this mod
1 near-identical copy found in the catalogue:
- zetty CLAUDE.md — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 1,446 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Zetty — Agent & Contributor Guide
Guidance for AI agents and contributors working in this repo.
CLAUDE.mdandAGENTS.mdare kept byte-identical — edit one and mirror the change to the other in the same commit (see Conventions).
What this is
Zetty is a native macOS (Linux later) GUI terminal multiplexer built on
full libghostty (via the prebuilt libghostty-spm package) with a Swift
AppKit application layer. Work is organized around pinnable projects, each
owning tabs and nested split panes. See README.md.
Layout
Sources/ZettyCore/**— pure, testable model (no AppKit):Surface,SurfaceNode,PaneTree,TabList,WorkspaceModel, persistence.App/Sources/App/**— AppKit app:AppDelegate,TerminalViewController,SidebarView,TabBarView,SurfaceNodeView,PaneActions,Theme.swift.App/Sources/ZettyGhostty/**— libghostty bridge:SurfaceRegistry,Ghostty.
Build / run
The Xcode project is Tuist-generated. Project.swift declares sources as
globs (App/Sources/App/**), but the generated .xcodeproj enumerates the
resolved file list — so after adding or removing a file you must regenerate.
You do not edit Project.swift to register a new file; only new targets,
resources, or settings need a manifest change. Files under Sources/ZettyCore/
need nothing at all for swift test (SwiftPM globs them at build time).
mise exec -- tuist generate --no-open
xcodebuild -project zetty.xcodeproj -scheme zetty -destination 'generic/platform=macOS' build
Tests: mise exec -- tuist test runs the app test target. The pure ZettyCore
suite is faster via SwiftPM — mise exec -- swift test — and a single test with
--filter, e.g. mise exec -- swift test --filter moveProjectRejectsCrossGroupMove.
Design rules ← read before any UI work
The visual spec (tokens, schemes, typography, component anatomy) is in
DESIGN.md; tokens live in
App/Sources/App/Theme.swift (ZTheme). DESIGN.md
is appearance-only — these enforceable rules (and Configuration, below) live here.
A change that violates one should be corrected before merge:
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.
- 2d ago First seen · 1,446 lines · 22,635 tokens per session scan D 0435438430de
zetty AGENTS.md is an instructions file published in the GitHub repository webteractive/zetty (2 stars, last pushed 5d ago), licensed MIT. It adds 22,635 tokens to every session, about $0.1132 per session on Opus 5. A static security scan graded it D with 3 findings (reads agent configuration directories, recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
liney AGENTS.md
Instructions for everettjf/liney, covering liney repository collaboration guide, 交流语言 / communication language, project overview, repository layout and build and test.
AppKitInspector AGENTS.md
AGENTS.md instructions for JaminZhou/AppKitInspector, covering appkit inspector — agent instructions, purpose, safety boundary, structure and verification.
liney CLAUDE.md
Instructions for everettjf/liney, a project described as: Liney is a native macOS terminal workspace built with AppKit, SwiftUI, and Ghostty, designed for multi-repo workflows, worktree switching, and focused vibe coding.
factoryfloor CLAUDE.md
Instructions for alltuner/factoryfloor, a project described as: AI-powered macOS development workspace. Git worktrees, Claude Code sessions, and dev servers in a single native app.
HomeClaw AGENTS.md
Instructions for omarshahine/HomeClaw, covering homeclaw, architecture, project structure, build system and xcode version.
skills-manager CLAUDE.md
Instructions for yibie/skills-manager, covering skills manager - development guide, tui engine: blessed, 核心原则, 参考文档 and 正确的 list 用法.