agentmods.dev

ToolNeuron: Instructions file for Claude Code

CLAUDE.md

ToolNeuron CLAUDE.md is an instructions file for Claude Code from Siddhesh2377/ToolNeuron. It costs 42,225 tokens per session, scanned A, original, MIT.

A project guide for ToolNeuron, an offline Android AI assistant that runs models and related features on the device.

In plain words
What is it for?
Use it when changing ToolNeuron's on-device AI, document search, voice, image generation, remote server, or plugin system.
Why use it?
It keeps future work aligned with the project's scope, privacy requirements, architecture, and current feature decisions.

Instructions file for Claude Code

Written for Claude Code: the file is CLAUDE.md. Also seen: mentions CLAUDE.md; mentions subagents.

This is Siddhesh2377/ToolNeuron's own configuration. It tells Claude Code how to work on ToolNeuron itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything ToolNeuron configures →

not rated 480repo +1 5mo ago A scan 42,225 tokens original MIT
View source ↗ Siddhesh2377/ToolNeuron
Reuse

Borrowing it

Nothing to install: this file belongs to Siddhesh2377/ToolNeuron. 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/Siddhesh2377/ToolNeuron/re-write/CLAUDE.md
Clone the repo
git clone --depth 1 https://github.com/Siddhesh2377/ToolNeuron

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.

agentmods badge for ToolNeuron CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/siddhesh2377/toolneuron/claude-md/github.svg)](https://agentmods.dev/instructions/siddhesh2377/toolneuron/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/siddhesh2377/toolneuron/claude-md"><img src="https://agentmods.dev/badge/instructions/siddhesh2377/toolneuron/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.

agentmods 80×15 button for ToolNeuron CLAUDE.md

Your own site · 80×15
<a href="https://agentmods.dev/instructions/siddhesh2377/toolneuron/claude-md"><img src="https://agentmods.dev/badge/instructions/siddhesh2377/toolneuron/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 42,225 This file is loaded in full into every session.
When invoked 42,225 The same file — it is already loaded in full.
Security scan A 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.42225 $0.42225
Opus 5.5 $0.16890 $0.16890
Sonnet 5.5 $0.08445 $0.08445
Haiku 4.5 $0.04222 $0.04222

Measured 2mo ago against content hash 48834f210825, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

ToolNeuron CLAUDE.md scanned grade A 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 2mo 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

`refreshModelCache()` hits `/v1/models` once per tab activation and filters per-kind for the model dropdowns. JNI: `nativeSetWebUiHtml(html)` pushes the bundled file at server start; `nativeClearWebUi()` clears on stop.
CLAUDE.md · 1,136 lines

How it starts

The opening of the file, as written. The whole thing — 1,136 lines — stays where its author put it; the contents beside it link to each section on GitHub.

ToolNeuron — Repo Guide

Project memory for this repo. When you change anything that affects future work — architecture, security behavior, new features, deprecated paths, public APIs, native JNI contracts, new scope — update this file as part of the same change. A future session reads this to reconstruct intent; if it drifts, work breaks.

Project scope

Privacy-first, offline-only on-device AI assistant. No Google Play services, no network telemetry, no analytics. In-scope pillars: on-device LLM chat, RAG over user documents, vision-language models (VLM), voice (TTS+STT), Remote Server with bundled web UI, HF Explorer, on-device image generation / img2img / inpaint / 4× upscale via the :ai_sd AAR (re-pivoted in 2026-05-08), first-party plugin system with ONNX inference + capability-gated APIs (re-pivoted in 2026-05-11). Out of scope: tool calling, Termux integration. (Image generation was originally cut on 2026-04-20 and re-added on 2026-05-08. Plugin marketplace was also originally cut on 2026-04-20 and re-added on 2026-05-11 as a first-party plugin runtime — DexClassLoader, Plugin contract with @Composable Content(), capability-gated OnnxApi/HxsApi/NetworkApi, floating plugin dock with smooth switch transitions.)

Package: com.dark.tool_neuron · minSdk 29 · targetSdk 36 · abiFilters arm64-v8a, x86_64.

Modules:

  • :app — UI (Compose), viewmodels, DI graph, activities, services.
  • :hxs — encrypted key-value store (Kotlin wrapper + C++ core).
  • :hxs_encryptor — crypto + integrity primitives: Argon2id, AES-GCM/ChaCha20-Poly1305, BoringSSL, ML-KEM-768, ML-DSA-65, Ed25519, HKDF, mmap+mlock SecureBuffer, plus the native security policy / auth / boot-integrity stack.
  • :native-server — embedded OpenAI-compatible HTTP server (cpp-httplib + nlohmann/json header-only via FetchContent, no BoringSSL / OpenSSL / zlib dep). Powers Remote Server mode.
  • :plugin-api — pure-Kotlin contract plugin authors compile against. Plugin interface with lifecycle + @Composable Content(), PluginContext, PluginCapability enum, PluginManifest, OnnxApi, HxsApi, NetworkApi. Compose deps are compileOnly; host provides them at runtime via classloader-parent.
  • :plugin-exc — plugin host runtime. PluginExecutor, PluginLoader (DexClassLoader from <filesDir>/plugins/<id>/classes.dex + .so detection across SUPPORTED_ABIS), PluginRegistry, PluginInstance, CapabilityGate (PolicyEngine.hasSession + manifest cap check), PluginContainerActivity (hosts plugin's Content() with AnimatedContent fade+scale plugin-switch transitions), PluginDock (floating M3-expressive surface with circle chips per open plugin, tertiary-corner native-code badge), OnnxApiImpl (wraps onnxruntime-android 1.20.0, gated by AI_ONNX), HxsApiImpl (per-plugin collection plugin_<id>, indexed string keys, gated by HXS_READ/WRITE), NetworkApiImpl (WebNative.fetchBytes GET, gated by INTERNET).
  • :download_manager, :networking — ancillary modules.

Read the full file on GitHub · 1,136 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. 2mo ago First seen · 1,136 lines · 42,225 tokens per session scan A 48834f210825

Subscribe to this mod's changes

ToolNeuron CLAUDE.md is an instructions file published in the GitHub repository Siddhesh2377/ToolNeuron (480 stars, last pushed 5mo ago), licensed MIT. It adds 42,225 tokens to every session, about $0.1689 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Mobileencryptedon-device-aiopen-sourceprivacy-first
Related

ToolNeuron CLAUDE.md compared with instructions from other repositories

ToolNeuron CLAUDE.md next to 6 instructions from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
ToolNeuron CLAUDE.mdSiddhesh2377/ToolNeuronInstructions file4805mo agoMITA42,225Claude Code
vscode buildNext.instructions.mdmicrosoft/vscodeInstructions file195kyesterdayMITA9,719GitHub Copilot
next.js AGENTS.mdvercel/next.jsInstructions file143kyesterdayMITA7,461Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchainInstructions file149kyesterdayMITA4,680Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscodeInstructions file195kyesterdayMITA5,516GitHub Copilot
vscode copilot-instructions.mdmicrosoft/vscodeInstructions file195kyesterdayMITA3,093GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harnessInstructions file274k2d agoMITA4,313Codex, OpenCode