What is jacky CLAUDE.md?

A project guide for Jack, a privacy-focused macOS voice assistant whose code package is called autobot. It describes the app’s local services, interface, speech features, and planning rules.

Use it when changing Jack’s chat interface, voice features, local Python service, macOS shell, or project planning process.

It gives coding agents and developers one place to understand the project structure and constraints, including which data may leave the computer.

How mdjakkariya/jacky configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to mdjakkariya/jacky. 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/mdjakkariya/jacky/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/mdjakkariya/jacky

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/mdjakkariya/jacky/claude-md.svg)](https://agentmods.dev/instructions/mdjakkariya/jacky/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/mdjakkariya/jacky/claude-md"><img src="https://agentmods.dev/badge/instructions/mdjakkariya/jacky/claude-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 2,821 This file is loaded in full into every session.

When invoked 2,821 The same file — it is already loaded in full.

Security scan A 0 findings. 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	$0.02821	$0.02821
Opus 5	$0.01411	$0.01411
Sonnet 5	$0.00564	$0.00564
Haiku 4.5	$0.00282	$0.00282

Measured 6d ago against content hash 4ea3a35e907c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

jacky 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 6d 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 · 191 lines

How it starts

The opening of the file, as written. The whole thing — 191 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

Guidelines for AI assistants (and humans) working in this repo — the single source of truth for how we build here. (For what's planned and its status, see GitHub Issues + Project #1, never markdown; this file holds no status.)

What this is

Jack — a local, privacy-first, English-only macOS voice assistant. (The codebase and Python package are named autobot; "Jack" is the product the user sees.) Everything runs on-device; no audio, text, or memory leaves the machine except the two disclosed, opt-in exceptions below.

The product is a floating orb plus a right-docked chat drawer (a Tauri shell over a system webview) that are thin clients of a headless Python daemon. Chat is the default; voice is opt-in, and the speech models download on demand the first time it's enabled (they're not bundled). The LLM is local (Ollama) by default, with an optional cloud provider (Anthropic).

Planning & tracking: GitHub Issues + Project #1 (see "Planning & tracking" below).
Architecture diagram: docs/architecture/architecture.svg; design reference: docs/architecture/design-reference.md.

Planning & tracking

Planning, feature requests, and status live in GitHub Issues + Project #1 ("Jack Assistent"), never in markdown. Do not create or edit tracking markdown (roadmaps, TODO docs, status checklists, "next steps" files) — that's what GitHub is for. To propose or plan work, open an issue (use the Feature request or Task template); it lands on the board automatically. To record how something works, add to docs/ (durable design reference: docs/architecture/design-reference.md). Every PR links its issue with Closes #NN.

Non-negotiable constraints

On-device only. Never add a dependency or call that sends audio, text, or user data off the machine. This is the entire point of the project. **The one sanctioned exceptions are opt-in, off by default, and disclosed: (a) the web_search tool (tools/web.py) sends only the search query and is only registered when allow_web is set; and (b) the optional cloud LLM (llm/anthropic_llm.py, llm_provider="anthropic") sends the conversation + memory profile + tool schemas/results to Anthropic — but never audio, and never the actions (those still run locally through the permission gate). Any other off-device feature needs the same explicit, opt-in, disclosed treatment. Secrets (API keys) live in the macOS Keychain (autobot.secrets), never on disk.
English only, both directions (STT and TTS). Prefer English-optimized models (Moonshine, Parakeet, *.en whisper builds). Do not reintroduce multilingual options.
The permission gate is not optional. Any genuinely-acting tool (write, delete, network, shell) must go through the registry's risk classification and the permission gate. Never let the LLM execute side effects unguarded.
Engine stays headless. UIs (orb, chat drawer) are thin clients of the daemon API; never build the assistant as a UI app.

Read the full file on GitHub · 191 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.

6d ago First seen · 191 lines · 2,821 tokens per session scan A 4ea3a35e907c

Subscribe to this mod's changes

jacky CLAUDE.md is an instructions file published in the GitHub repository mdjakkariya/jacky (2 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 2,821 tokens to every session, about $0.0141 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.

ai-agents chat-assistant macos voice-assistant

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 · 5,182 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

jacky: Instructions file for Claude Code