PlamenTSV/plamen

Autonomous Web3 security audit agent for Claude Code

What it solves

Plamen is an autonomous security auditor designed to find vulnerabilities in smart contracts and Layer 1 (L1) node-client infrastructure. It automates the complex process of auditing code across multiple blockchain ecosystems, reducing the need for manual security reviews by producing verified reports with proof-of-concept (PoC) exploits.

How it works

Plamen orchestrates a swarm of 18 to 100 AI agents across 40+ analysis phases. It integrates with AI runtimes like Claude Code and OpenAI Codex CLI to execute its workflow. The system uses a combination of static analysis (via MCP servers like Slither), a custom RAG (Retrieval-Augmented Generation) vulnerability database, and executable PoC attempts to verify findings. It supports a wide range of languages and platforms, including EVM/Solidity, Solana/Anchor, Aptos/Sui Move, and Go/Rust for L1 infrastructure.

Who it’s for

Security researchers, blockchain developers, and auditors who need automated, high-depth security analysis for smart contracts or blockchain node clients.

Highlights

  • Multi-Chain Support: Works across EVM, Solana, Aptos, Sui, Soroban, DAML, and L1 Go/Rust clients.
  • Verified Findings: Forces executable proof-of-concept (PoC) attempts for material harms to eliminate false positives.
  • Tiered Audit Modes: Offers Light, Core, and Thorough modes to balance speed, cost, and depth of analysis.
  • L1 Infrastructure Focus: Includes specialized skills for auditing consensus engines, p2p networking, and validator lifecycle code.
  • RAG Integration: Features an optional vulnerability database to ground AI agents in known security patterns.

Related

  • Project
    NX-AI/tirex

    A pre-trained 35M parameter time series model based on xLSTM that provides zero-shot forecasting, classification, and regression across long and short horizons.

  • Project
    Mengqi97/chinese-medical-dataset

    A curated collection of open-source Chinese medical datasets for tasks like medical QA, entity recognition, and knowledge graph construction to support the development of medical AI.

  • Project
    TornLux/UnrealBridge

    A bridging layer that enables AI agents to control and edit Unreal Engine 5.3+ in real time, featuring AST-based validation to prevent hallucinations and deep integration with Blueprints, VFX, and rigging.

  • Project
    Unagi-cq/cdp-bridge-mcp

    A bridge service that connects MCP clients to real browser sessions, allowing AI agents to read, scan, and automate actions in a user's already-logged-in browser pages.