RustFox CLAUDE.md

Development instructions for RustFox, a Telegram bot written in Rust that uses an AI model and sandboxed tools. They explain the project, configuration, and commands for building, running, formatting, checking, and linting it.

In plain words
What is it for?
Use them when developing, building, testing, formatting, or linting RustFox, including when choosing a custom configuration file or setting up its required credentials.
Why use it?
They give an unfamiliar developer the steps and settings needed to work on the project without guessing how it is structured or run.

Instructions file

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 instructions/chinkan/rustfox/claude-md
Clone the repo
git clone --depth 1 https://github.com/chinkan/RustFox
Per session 4,170 This file is loaded in full into every session.
When invoked 4,170 The same file — it is already loaded in full.
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 $0.04170 $0.04170
Opus 5 $0.02085 $0.02085
Sonnet 5 $0.00834 $0.00834
Haiku 4.5 $0.00417 $0.00417

Measured 2d ago against content hash 22b22dbcac6d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

RustFox CLAUDE.md 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.md · 373 lines

How it starts

The opening of the file, as written. The whole thing — 373 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md - RustFox Development Guide

Project Overview

RustFox is a Telegram AI assistant written in Rust. It connects to Telegram as a bot, uses OpenRouter LLM for inference (default model: qwen/qwen3-235b-a22b), provides built-in sandboxed tools (file I/O, command execution), and supports MCP (Model Context Protocol) servers for extensible tool integration. It implements an agentic loop that iterates tool calls until a final text response is produced (max iterations configurable, default 25).

Build & Run

# Build (debug)
cargo build

# Build (release)
cargo build --release

# Run (uses ./config.toml by default)
cargo run

# Run with custom config path
cargo run -- /path/to/config.toml

# Check without building
cargo check

# Format code
cargo fmt

# Lint
cargo clippy

Configuration

Copy config.example.toml to config.toml and fill in credentials. The config.toml file is gitignored and must never be committed. Required fields:

  • telegram.bot_token - Telegram Bot API token
  • telegram.allowed_user_ids - Whitelist of Telegram user IDs
  • openrouter.api_key - OpenRouter API key
  • sandbox.allowed_directory - Directory for sandboxed file/command operations

Home directory

RustFox stores all state under a single home directory (default ~/.rustfox), resolved as: RUSTFOX_HOME env (absolute) → [general].home config → ~/.rustfox. Layout: config.toml, rustfox.db, skills/, agents/, workspace/ (the sandbox), artifacts/, user_model.md. Each path can be pinned to an absolute location in config.toml; unset paths fall back to the home default. Run isolated instances with RUSTFOX_HOME=.... See docs/persistent-home-directory.md. Path resolution lives in src/home.rs (Config::resolve writes the resolved absolute paths back into the config). Bundled skills/agents are seed-copied on first run; /update-skills re-syncs them using <home>/skills-lock.json.

Architecture

src/
├── main.rs      # Entry point: logging init, config loading, MCP setup, bot launch
├── config.rs    # TOML config parsing (Config, TelegramConfig, OpenRouterConfig, SandboxConfig, McpServerConfig)
├── llm.rs       # OpenRouter API client (ChatMessage, ToolCall, ToolDefinition, LlmClient)
├── tools.rs     # Built-in tool definitions and execution with sandbox path validation
├── mcp.rs       # MCP client manager (McpManager, McpConnection) for external tool servers
└── bot.rs       # Telegram bot handler: message routing, agentic loop, conversation state

Read the full file on GitHub · 373 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 · 373 lines · 4,170 tokens per session scan A 22b22dbcac6d

Subscribe to this mod's changes

RustFox CLAUDE.md is an instructions file published in the GitHub repository chinkan/RustFox (7 stars, last pushed 26d ago), licensed MIT. It adds 4,170 tokens to every session, about $0.0209 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 2026-08-31.