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/mpolatcan/codehub/claude-mdgit clone --depth 1 https://github.com/mpolatcan/codehubWhat 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 | $0.08719 | $0.08719 |
| Opus 5 | $0.04359 | $0.04359 |
| Sonnet 5 | $0.01744 | $0.01744 |
| Haiku 4.5 | $0.00872 | $0.00872 |
Grade C, and why
codehub CLAUDE.md scanned grade C with 2 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.
Downloads and executes remote codehighSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
- **`SHELL ["bash", "-euo", "pipefail", "-c"]` is set in the runtime Dockerfile.** This matters because we use `curl ... | bash` patterns to install CLIs; without pipefail, a failing curl silently succeeds and the CLI is Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **`SHELL ["bash", "-euo", "pipefail", "-c"]` is set in the runtime Dockerfile.** This matters because we use `curl ... | bash` patterns to install CLIs; without pipefail, a failing curl silently succeeds and the CLI is How it starts
The opening of the file, as written. The whole thing — 218 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CodeHub — Repository Guide for Claude
A Tauri 2 desktop app that runs multiple AI coding CLIs (Claude Code, Codex, Antigravity) in Docker containers, multiplexed via tmux. Each tab in the window = one tmux session = one agent. Each workspace gets its own container (codehub-ws-<key>) so the hub's fleet view reports REAL per-workspace cpu/mem/net/state.
This file is loaded on every Claude session in this repo. Keep it short and load-bearing.
Behavioral rules
- Think before coding. State assumptions explicitly. If multiple interpretations exist, present them — don't pick silently. If something is unclear, stop and ask. Push back when a simpler approach exists.
- Simplicity first. Minimum code that solves the problem. No features beyond what was asked. No abstractions for single-use code. No speculative "flexibility". If 200 lines could be 50, rewrite it.
- Surgical changes. Touch only what you must. Don't "improve" adjacent code, comments, or formatting. Match existing style. Remove imports/variables YOUR changes made unused — don't remove pre-existing dead code unless asked. Every changed line should trace to the user's request.
- Goal-driven execution. Transform tasks into verifiable goals. For multi-step tasks, state a brief plan with checkpoints. Build → verify → iterate, not build-everything-then-check.
Stack
- Backend: Rust 1.85+, Tauri 2, tokio, bollard (Docker API client), futures-util.
- Frontend: React 19 + TypeScript, Vite 5, Zustand 5 (state), Tailwind v4 + shadcn/Radix (Dialog + Popover) for chrome, xterm.js 5 (+ fit + canvas addons) for the terminal panes. xterm panes live OUTSIDE React in a non-reactive registry (
src/app/lib/panes.ts) —<PaneMount>reparents the DOM node so buffers survive splits and tab switches. - Runtime image: Debian-based
node:20-slimbase,tail -f /dev/nullentrypoint, tmux server spawned on demand. - OS support: macOS primary, Linux secondary, Windows untested.
Repo layout
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 · 218 lines · 8,719 tokens per session scan C 8677757d5d30
codehub CLAUDE.md is an instructions file published in the GitHub repository mpolatcan/codehub (5 stars, last pushed 1mo ago), licensed MIT. It adds 8,719 tokens to every session, about $0.0436 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, 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
ephemeral-sandbox CLAUDE.md
Instructions for Ephemeral-AI-Lab/ephemeral-sandbox, covering claude.md, project, engineering practice (required), build & test and sandbox tools.
ephemeral-sandbox AGENTS.md
Instructions for Ephemeral-AI-Lab/ephemeral-sandbox, covering agents.md and required rules.
clawpier CLAUDE.md
Instructions for SebastianElvis/clawpier, covering claude.md, project overview, tech stack, development commands and running single tests.
seemux CLAUDE.md
Claude Code instructions for asermax/seemux, covering claude.md, project overview, build & development commands, architecture and data flow.
appwrite AGENTS.md
Instructions for appwrite/appwrite, covering appwrite, commands, stack, layout and libraries.
netdata AGENTS.md
AGENTS.md instructions for netdata/netdata, covering agents.md, goals, requirement language, mandatory development principles and sow system.