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/karvidsson/agent-ptt/claude-mdgit clone --depth 1 https://github.com/karvidsson/agent-pttWrote 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/karvidsson/agent-ptt/claude-md)<a href="https://agentmods.dev/instructions/karvidsson/agent-ptt/claude-md"><img src="https://agentmods.dev/badge/instructions/karvidsson/agent-ptt/claude-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 5.1 | $0.01166 | $0.01166 |
| Opus 5 | $0.00583 | $0.00583 |
| Sonnet 5 | $0.00233 | $0.00233 |
| Haiku 4.5 | $0.00117 | $0.00117 |
Grade A, and why
agent-ptt 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 5d 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 — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Project
Agent PTT — voice channels for AI agents. Participants (agents/humans) join named channels via CLI or WebSocket and say text messages; the server synthesizes them to speech, plays them through the host's speakers, and streams the audio to WebSocket spectators. Python 3.11+, managed with uv.
Commands
uv sync # install dependencies (base: no torch)
uv sync --extra omnivoice # + local neural TTS (torch; model downloads ~2.4 GB on first use)
uv run agent-ptt server start # start the FastAPI server (default port 8770)
uv run agent-ptt channel create "Name" # create a channel
uv run agent-ptt join <channel-id> --handle "Name" --voice "en-US-GuyNeural"
uv run agent-ptt say "text" # speak into the joined channel (plays on speakers)
uv run agent-ptt listen <channel-id> # spectate a channel's audio stream
uv run agent-ptt voices # list available TTS voices
uv run agent-ptt config # show current session state
uv run pytest # run the test suite
uv run pytest tests/test_api.py -k name # run a single test file / test
uv run ruff check . # lint (auto-fix with --fix)
uv run ruff format . # format
Run pytest and ruff before committing. Tests fake out TTS and speaker playback (see tests/conftest.py: FakeTTSBackend, FakeMixer) and use a temp SQLite DB, so they're fast and need no network or audio hardware. The audible end-to-end path (real TTS → speakers) is only verifiable manually — see docs/testing.md. CI (.github/workflows/ci.yml) runs the same gates plus scripts/validate_plugins.py; releases are cut by tagging vX.Y.Z after bumping every plugins/*/.claude-plugin/plugin.json version to match (see plugins/README.md).
Architecture
Single package agent_ptt/, three layers: Typer CLI client → FastAPI server → services (channel manager, TTS, audio mixer, DB).
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.
- 5d ago First seen · 52 lines · 1,166 tokens per session scan A 460c6e9e1167
agent-ptt CLAUDE.md is an instructions file published in the GitHub repository karvidsson/agent-ptt (2 stars, last pushed 1mo ago), licensed MIT. It adds 1,166 tokens to every session, about $0.0058 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
cladding CLAUDE.md
Claude Code instructions for qwerfunch/cladding, covering maintainer guide for ai-assisted work on cladding, spec authoring — always hash, never f-nnn, version bumps — use the script, detector additions — count is auto-recomputed and plugin manifest mirrors — built, not hand-authored.
cladding AGENTS.md
AGENTS.md instructions for qwerfunch/cladding, covering agents.md, 1. project, 2. setup, 3. verify before pushing and 4. code & comment style.
sigil CLAUDE.md
Instructions for Anmol-Srv/sigil, covering sigil and design system.
bounded-relay AGENTS.md
AGENTS.md instructions for mohammad19974/bounded-relay, covering repository operating rules, product boundary, before changing code, implementation invariants and quality gates.
cortex-viz CLAUDE.md
Claude Code instructions for cdeust/cortex-viz, covering cortex-viz, repo-specific constraints and etiquette.
grpvn GEMINI.md
Instructions for frane/grpvn, covering grpvn technical instructions, design, routing and output.