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/maxanatsko/agentic-voice/agents-mdgit clone --depth 1 https://github.com/maxanatsko/agentic-voiceWrote 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.
[](https://agentmods.dev/instructions/maxanatsko/agentic-voice/agents-md)<a href="https://agentmods.dev/instructions/maxanatsko/agentic-voice/agents-md"><img src="https://agentmods.dev/badge/instructions/maxanatsko/agentic-voice/agents-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00590 | $0.00590 |
| Opus 5 | $0.00295 | $0.00295 |
| Sonnet 5 | $0.00118 | $0.00118 |
| Haiku 4.5 | $0.00059 | $0.00059 |
Grade A, and why
agentic-voice AGENTS.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 3d 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.
How it starts
The opening of the file, as written. The whole thing — 63 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Guidelines
Purpose
This repository provides a minimal local voice-output capability for coding agents, speaking through a local TTS backend over a fixed HTTP contract. The bundled backend (bridge/) wraps Kokoro-82M via FluidAudio; it is not an NVIDIA product.
Engineering principles
Separation of concerns
Keep boundaries explicit:
- MCP code handles MCP concerns only.
- CLI code handles process input/output only.
VoiceServiceowns the speak use case and user-facing speech policy.- TTS backend HTTP details stay inside the TTS client (
src/tts/tts-client.ts). - Audio playback stays inside the audio module.
- Configuration parsing stays in
config.ts. bridge/is a separate Swift package with its own toolchain; Node code must depend on it only through the documented HTTP contract, never by assuming its internals.
Do not let transport-specific objects leak into application code.
KISS
Prefer direct code over patterns. A small function or class is better than a framework, registry, factory hierarchy, or dependency injection container when there is only one implementation.
YAGNI
Do not add capabilities without a concrete requirement. In particular, avoid adding:
- provider abstraction layers before a second TTS provider exists;
- persistence, queues, retries, telemetry, or a web UI;
- an embedded summarization LLM, including inside a hook — hooks (
hooks/) stay fixed-phrase and deterministic, never transcript-parsing; - speech-to-text or full-duplex audio.
DRY
Share behavior, not coincidental syntax. MCP and CLI must use the same VoiceService. There should be one implementation each for TTS synthesis, speech-length policy, configuration, and audio playback.
Do not create generic helpers solely to remove two or three similar lines.
Code conventions
- TypeScript, ESM, strict mode.
- Prefer the Node.js standard library over dependencies.
- Validate external input at boundaries.
- Keep functions small and intention-revealing.
- Fail with actionable error messages; do not silently swallow synthesis or playback failures.
- Never write diagnostic output to stdout from the MCP process. stdout is reserved for MCP protocol traffic; use stderr for diagnostics.
- Tests should focus on deterministic application behavior. Do not mock implementation details merely to increase coverage.
- Swift code in
bridge/is not built or tested by the roottsc/node --testpipeline — verify it withswift build --package-path bridge(seebridge/README.md).
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.
- 3d ago First seen · 63 lines · 590 tokens per session scan A 12daa0f452ba
agentic-voice AGENTS.md is an instructions file published in the GitHub repository maxanatsko/agentic-voice (1 stars, last pushed 9d ago), licensed MIT. It adds 590 tokens to every session, about $0.0030 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.
Other instructions, from other repositories
parallel-code CLAUDE.md
Instructions for johannesjo/parallel-code, covering parallel code, stack, commands, project structure and conventions.
cezar AGENTS.md
Instructions for open-mercato/cezar, covering agents.md — working in this repository, zero config, changing a mechanism that already works, the http api and repository layout.
many-ai-cli CLAUDE.md
Instructions for ishizakahiroshi/many-ai-cli, covering many-ai-cli 開発ガイド, プロジェクト概要, 新規 provider を増やす提案をしない(2026-08-13 制定), デスクトップアプリ化・microsoft store 提出は提案しない(2026-08-14 制定) and 現在の実装状態.
agent-audit AGENTS.md
Instructions for fullstackdegen/agent-audit, covering agent audit instructions, core rules, applying fix packs, what this tool covers and completion standard.
agent-audit CLAUDE.md
Instructions for fullstackdegen/agent-audit, covering claude code instructions, mcp setup, how to use reports, suggested claude prompt and working in this repository.
seamless AGENTS.md
Instructions for 0spoon/seamless, covering agents.md, go style rules, general, package structure and naming.