What is siphon-ai CLAUDE.md?

Repository instructions for SiphonAI, a Rust program that connects telephone systems to a developer's WebSocket server. It carries call audio between them, while the developer's server provides any AI processing.

It helps agents understand the project's architecture, workspace layout and boundaries when changing SIP calls, audio streaming or related code.

It prevents coding in the wrong part of the system, such as adding AI-provider calls to the phone-and-audio bridge.

How thevoiceguy/siphon-ai configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to thevoiceguy/siphon-ai. 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/thevoiceguy/siphon-ai/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/thevoiceguy/siphon-ai

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/thevoiceguy/siphon-ai/claude-md/github.svg)](https://agentmods.dev/instructions/thevoiceguy/siphon-ai/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/thevoiceguy/siphon-ai/claude-md"><img src="https://agentmods.dev/badge/instructions/thevoiceguy/siphon-ai/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.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/thevoiceguy/siphon-ai/claude-md"><img src="https://agentmods.dev/badge/instructions/thevoiceguy/siphon-ai/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 6,842 This file is loaded in full into every session.

When invoked 6,842 The same file — it is already loaded in full.

Security scan C 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.

Model	Per session	Once invoked
Fable 5.1	$0.06842	$0.06842
Opus 5.5	$0.02737	$0.02737
Sonnet 5.5	$0.01368	$0.01368
Haiku 4.5	$0.00684	$0.00684

Measured 2mo ago against content hash d3a06557155e, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade C, and why

siphon-ai CLAUDE.md scanned grade C 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.

Tells the agent never to refusehighAnti-refusal

Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.

**Delivered since the original v1 cut** (no longer out of scope — these shipped and have their own docs/config): call recording (0.5.0), outbound origination + attended transfer (0.6.0/0.6.1), conferencing/mixing and med

CLAUDE.md · 485 lines

How it starts

The opening of the file, as written. The whole thing — 485 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

Operating instructions for AI coding agents (Claude Code, Cursor, etc.) working in this repo. Read this file first, before any task. Re-check relevant sections before non-trivial changes.

For the what and why of this project, see docs/DEV_PLAN.md. This file is the how.

1. What SiphonAI Is (One Paragraph)

SiphonAI is a SIP-to-WebSocket media bridge written in Rust. It accepts inbound SIP calls (either as a trunk endpoint or as a registered phone on a PBX), streams the call's audio over a WebSocket to a developer-supplied server, and plays audio received back over that WebSocket into the call. It does not contain any AI code. AI is the WebSocket server's job. SiphonAI's job is the bridge: SIP signaling, RTP media, codec handling, jitter, barge-in, DTMF, hold/transfer — all wrapped in a clean WS protocol.

If you find yourself writing code that calls an AI provider (OpenAI, Anthropic, Deepgram, ElevenLabs, etc.), stop. That's the wrong layer. Ask the user.

2. Architecture in 60 Seconds

PBX/Trunk ──SIP/RTP──► SiphonAI ──WebSocket──► Developer's WS server (BYO AI)
                          │
                          ├── siphon-rs        (SIP stack — external dep)
                          ├── forge-media      (RTP, codecs, SDP — external dep)
                          └── our code         (orchestration + WS protocol)

Two upstream Rust libraries do most of the heavy lifting:

siphon-rs (https://github.com/thevoiceguy/siphon-rs) — RFC 3261 SIP stack. UAS, UAC, transactions, dialogs, REGISTER, REFER, auth.
forge-media (https://github.com/thevoiceguy/forge-media) — RTP/RTCP, codecs (G.711/Opus), SDP (forge-sdp), jitter buffer, DTMF, audio injection, VAD.
hep-rs (new, owned by us) — HEP3 codec, transport, HepSink trait. Used by all three of siphon-rs (SIP messages), forge-media (RTCP/QoS), and siphon-ai (logs/CDRs/events) to ship observability data to Homer.

All three are owned by the same author as SiphonAI. If you find a missing capability in any, the right answer is often a small PR upstream, not a workaround here. Ask the user before doing this.

Read the full file on GitHub · 485 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.

2mo ago Changed · +4 lines · +81 tokens per session d3a06557155e
2mo ago First seen · 481 lines · 6,761 tokens per session scan C e8d315588438

Subscribe to this mod's changes

siphon-ai CLAUDE.md is an instructions file published in the GitHub repository thevoiceguy/siphon-ai (34 stars, last pushed 14d ago), licensed Apache-2.0. It adds 6,842 tokens to every session, about $0.0274 per session on Opus 5.5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Languages Planning Rust call-recording calls hep homer rtp

siphon-ai CLAUDE.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
siphon-ai CLAUDE.mdthevoiceguy/siphon-ai	Instructions file	34	14d ago	Apache-2.0	C	6,842	Claude Code
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	9,719	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	yesterday	MIT	A	7,461	Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	yesterday	MIT	A	4,680	Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	5,516	GitHub Copilot
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	3,093	GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness	Instructions file	274k	2d ago	MIT	A	4,313	Codex, OpenCode

siphon-ai: Instructions file for Claude Code