core

A set of rules describing a shared speech-to-text and text-to-speech layer for a Rust web server. Speech-to-text turns audio into words, while text-to-speech turns written words into audio, across several providers.

In plain words
What is it for?
Connecting to speech providers, sending audio for transcription, producing spoken audio, and managing provider connections through a common Rust interface.
Why use it?
It gives different speech services one common WebSocket interface, so the rest of the application does not need separate provider-specific handling. It also defines the expected connection and audio methods.

Cursor rule for Cursor

Install

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.

agentmods
npx agentmods add rules/saynaai/sayna/core
Clone the repo
git clone --depth 1 https://github.com/SaynaAI/sayna

Made for: Cursor.

Per session 1,582 This file is loaded in full into every session.
When invoked 1,582 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
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 $0.01582 $0.01582
Opus 5 $0.00791 $0.00791
Sonnet 5 $0.00316 $0.00316
Haiku 4.5 $0.00158 $0.00158

Measured 2d ago against content hash cd195b9d747e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

core 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 2d 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.

.cursor/rules/core.mdc · 131 lines

How it starts

The opening of the file, as written. The whole thing — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Sayna Core Module

This Rust Axum project is a unified STT, TTS server that handles real-time WebSocket APIs for STT and TTS providers like Deepgram, ElevenLabs, Google, Microsoft Azure, Cartesia, and others into one unified WebSocket API that abstracts away provider-specific details.

The main business logic is inside src/core/, which uses trait-based abstraction to have unified higher-level implementations of TTS and STT separated into dedicated folders.

STT Base Abstraction (src/core/stt/base.rs)

The BaseSTT trait defines the interface for all STT providers:

Required Methods

Method Signature Description
new fn new(config: STTConfig) -> Result<Self, STTError> Create a new instance with STT configuration
connect async fn connect(&mut self) -> Result<(), STTError> Initiate connection to STT provider
disconnect async fn disconnect(&mut self) -> Result<(), STTError> Disconnect from provider
is_ready fn is_ready(&self) -> bool Check if connection is ready
send_audio async fn send_audio(&mut self, audio_data: Vec<u8>) -> Result<(), STTError> Send audio bytes for transcription
on_result async fn on_result(&mut self, callback: STTResultCallback) -> Result<(), STTError> Register transcription result callback
on_error async fn on_error(&mut self, callback: STTErrorCallback) -> Result<(), STTError> Register error callback for streaming errors
get_config fn get_config(&self) -> Option<&STTConfig> Get current configuration
update_config async fn update_config(&mut self, config: STTConfig) -> Result<(), STTError> Update configuration while connected
get_provider_info fn get_provider_info(&self) -> &'static str Get provider-specific information

STTResult Type

pub struct STTResult {
    pub transcript: String,      // The transcribed text
    pub is_final: bool,          // Whether this is a final result
    pub is_speech_final: bool,   // Whether speech segment ended
    pub confidence: f32,         // Confidence score (0.0 to 1.0)
}

Read the full file on GitHub · 131 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. 2d ago First seen · 131 lines · 1,582 tokens per session scan A cd195b9d747e

Subscribe to this mod's changes

core is a cursor rule published in the GitHub repository SaynaAI/sayna (311 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 1,582 tokens to every session, about $0.0079 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 2026-08-30.