Sol Shred SDK - Solana ShredStream & Multi-DEX Event Parser

sol-shred-sdk is a low-latency Rust SDK for Solana raw shred decoding, ShredStream ingestion, and multi-DEX transaction event parsing. It turns Solana UDP shreds or Jito-style ShredStream entries into typed DexEvent data for trading bots, snipers, indexers, and real-time analytics.

The default, lowest-latency path is:

UDP packet -> Solana/Agave Shred -> FEC recovery -> deshred -> Entry -> VersionedTransaction -> event parser

Source/decode mode is selected with ShredDecodeMode. Native raw UDP shreds are the recommended path; Jito-style ShredStream gRPC entries are retained as a compatibility source while Jito service availability lasts.

What This SDK Provides

Area	Coverage
Input	Raw UDP Solana shred payloads, or Jito-style ShredStream gRPC entries
Decode	ShredDecodeMode::RawUdp uses solana-ledger Shred parsing, Reed-Solomon recovery, Shredder::deshred, and bincode Vec<Entry> decode; ShredDecodeMode::JitoGrpc receives prebuilt entry batches
Transactions	Entry-to-transaction flattening with slot context
Events	DexEvent parser migrated from sol-parser-sdk ShredStream handling
Extensibility	TransactionEventParser trait for custom parser plug-ins
Compatibility	Preserves selected generated proto/type re-exports used by older code

Migrated parser families match the sol-parser-sdk parser surface:

• PumpFun, PumpFun v2/Mayhem, Pump fees, PumpSwap
• Raydium LaunchLab, CPMM, CLMM, AMM V4
• Orca Whirlpool
• Meteora Pools, DAMM V2, DBC, DLMM
• Token accounts, nonce accounts, selected DEX account state events, and block metadata types

Use Cases

• Solana low-latency trading bots and sniper bots
• Real-time PumpFun, PumpSwap, Raydium, Orca, and Meteora event feeds
• Raw shred UDP ingestion pipelines without relying on Jito ShredStream
• Jito ShredStream gRPC compatibility while migrating to raw shreds
• Multi-DEX indexers, analytics, and alerting systems
• Higher-level parser SDKs that want to reuse one shred/entry ingestion layer

Raw shred subscriptions parse transaction-visible instruction data directly from Entry transactions. Log-only and account-update event parsers are included for SDK compatibility, but raw shreds do not carry execution logs or account update payloads by themselves.

Installation

[dependencies]
sol-shred-sdk = "3.0.1"

Decode Mode

Choose the source/decoder when creating the client:

use sol_shred_sdk::{RawShredConfig, ShredDecodeMode, ShredStreamClient};

let raw = ShredStreamClient::new_with_decode_mode(
    ShredDecodeMode::raw_udp(RawShredConfig::default()),
).await?;

let jito = ShredStreamClient::new_with_decode_mode(
    ShredDecodeMode::jito_grpc("http://127.0.0.1:10000"),
).await?;

Raw UDP Event Subscription

Bind a UDP socket where your Solana shred source sends raw shred datagrams:

use sol_shred_sdk::shredstream::{ShredStreamClient, ShredStreamConfig};
use sol_shred_sdk::{DexEvent, EventType, EventTypeFilter};

#[tokio::main]
async fn main() -> anyhow::Result<()> {
    let config = ShredStreamConfig::low_latency()
        .with_udp_bind("0.0.0.0:8001".parse()?);

    let client = ShredStreamClient::new_with_config(config).await?;
    client
        .subscribe_with_filter_callback(
            Some(EventTypeFilter::include_only(vec![
                EventType::PumpFunCreate,
                EventType::RaydiumCpmmSwap,
                EventType::OrcaWhirlpoolSwap,
            ])),
            |event| {
                match event {
                    DexEvent::PumpFunCreate(create) => println!("pump create: {create:?}"),
                    DexEvent::RaydiumCpmmSwap(swap) => println!("cpmm swap: {swap:?}"),
                    DexEvent::OrcaWhirlpoolSwap(swap) => println!("orca swap: {swap:?}"),
                    other => println!("{other:?}"),
                }
            },
        )
        .await?;

    tokio::signal::ctrl_c().await?;
    client.stop().await;
    Ok(())
}

Queue Subscription

For consumers that prefer polling:

use sol_shred_sdk::shredstream::ShredStreamClient;

#[tokio::main]
async fn main() -> anyhow::Result<()> {
    let client = ShredStreamClient::new("0.0.0.0:8001").await?;
    let queue = client.subscribe().await?;

    loop {
        while let Some(event) = queue.pop() {
            println!("{event:?}");
        }
        tokio::task::yield_now().await;
    }
}

For legacy PumpFun/Bonk-only consumers, subscribe_pumpfun, subscribe_pumpfun_callback, and the generic subscribe_with_parser APIs are still available.

Low-Level Decoder

Use the low-level API when you already own the UDP receive loop:

use std::time::Instant;
use sol_shred_sdk::{RawShredConfig, RawShredDecoder};

fn handle_packet(decoder: &mut RawShredDecoder, packet: &[u8]) {
    for batch in decoder.push_packet(packet, Instant::now()) {
        for entry in batch.entries {
            for transaction in entry.transactions {
                println!("slot={} sig={:?}", batch.slot, transaction.signatures.first());
            }
        }
    }
}

let mut decoder = RawShredDecoder::new(RawShredConfig::default());

Custom Transaction Parser

sol-shred-sdk owns networking, shred reassembly, and transaction traversal. Higher-level crates can plug in event parsing without reimplementing shredstream handling:

use sol_shred_sdk::common::AnyResult;
use sol_shred_sdk::parser::TransactionEventParser;
use solana_sdk::transaction::VersionedTransaction;

struct MyParser;

impl TransactionEventParser for MyParser {
    type Event = String;

    fn parse_transaction_events<F>(
        &mut self,
        transaction: &VersionedTransaction,
        slot: u64,
        tx_index: u64,
        recv_us: i64,
        mut emit: F,
    ) -> AnyResult<usize>
    where
        F: FnMut(Self::Event),
    {
        if let Some(signature) = transaction.signatures.first() {
            emit(format!("slot={slot} tx_index={tx_index} recv_us={recv_us} sig={signature}"));
            return Ok(1);
        }
        Ok(0)
    }
}

Configuration Notes

• RawShredConfig::udp_payload_prefix_skip should stay 0 for native raw shreds.
• RawShredConfig::forward_slot_watermark defaults to false so out-of-order completed slots are not dropped. Enable it only when you explicitly prefer forward-only latency over completeness.
• ShredStreamConfig::low_latency() favors shorter reassembly waits and faster restart.
• ShredStreamConfig::high_throughput() increases receive buffering, tracked slots, and queue capacity.
• ShredDecodeMode::JitoGrpc keeps the old entries-gRPC source but still uses the same unified DexEvent parser.
• The UDP receive buffer is requested with socket2; the OS may cap the actual value.

Decoder Benchmark

The raw decoder has an ignored release-mode microbench that generates real Solana merkle shreds with solana-ledger and decodes them through RawShredDecoder:

RAW_SHRED_BENCH_ITERS=10000 cargo test --release --lib bench_decode_generated_shreds -- --ignored --nocapture

Current local reference result:

packets_per_sec=1172196 slots_per_sec=36631 tx_per_sec=1172196

License

This project is licensed under the MIT License - see the LICENSE file for details.

Telegram group

https://t.me/fnzero_group