| Repository | Focus | Capabilities |
|---|---|---|
rollback |
Agent Tool Rollback Engine | Transactional filesystem & state rollback engine for autonomous AI coding agents — <2ms micro-checkpoints, atomic reverts, and unified diff inspection. |
solana-tx-lens |
Solana Simulation Inspector | Headless transaction simulation inspector, recursive CPI tree visualizer, and Anchor custom error decoder for Solana. |
agent-witness |
Cryptographic Action Attestation | RFC 6962 Merkle Action Trees & verifiable .witness audit receipts for autonomous AI agents, verified on-chain via Solidity. |
agent-canary |
Zero-Knowledge Canaries | Cryptographic honeytokens and taint tripwires for autonomous AI agents — zero false-positive prompt injection detection. |
polymarket-depth |
CLOB Depth & Arbitrage | Real-time L2/L3 orderbook depth reconstructor and cross-outcome probability arbitrage scanner for Polymarket. |
solana-jito-core |
Jito MEV Bundler SDK | Lightweight Python SDK for building, simulating, and broadcasting sandwich-proof atomic transaction bundles. |
mcp-shield |
MCP Security Gateway | Real-time scanner for prompt injection, secret exfiltration, and unauthorized tool calls across Model Context Protocol servers. |
tripwire |
Agent Cost Circuit Breaker | Prevents runaway autonomous loops and token budget burn before cloud bill shock. |
jev-polymarket-executor |
Polymarket CLOB Engine | Automated probability mispricing detection, orderbook streaming, and paper/live trading execution. |
crashpad |
DeFi Safety Rails | Pre-flight transaction simulation, MEV sandwich guard, and dynamic slippage bounds for Web3 agents. |
hydra |
x402 Payment Router | Self-hostable HTTP-402 payment facilitator & multi-failover routing gateway for agent microtransactions. |
cerberus |
Payment Firewall | Blocks transaction hijacking, unauthorized spends, and wallet drainers before signing. |
crucible |
Continuous Agent Evals | Captures real production agent failures and automatically compiles regression test suites. |
jev-hyperliquid |
Hyperliquid L1 Engine | Sub-second WebSocket orderbook streaming, limit/market execution, and position liquidation risk manager. |
cognitive-automaton-bench |
Cognition Benchmark | Benchmarking emergent cognition and self-replicating behavior in LLMs inspired by Conway's Game of Life. |
⭐ Star any repository you find useful to bookmark it | Follow @devsoniclk for weekly production tools & infrastructure releases.



