Skip to content

Complementary MCP: 27K real audit findings as empirical signal for Plamen's agent workflow #20

Description

@holistis

Hey Plamen team,

I noticed Plamen is an autonomous Web3 security audit agent for Claude Code. I built a complementary MCP tool that adds empirical data to the audit workflow.

Bug Bounty Intelligence: https://github.com/holistis/bug-bounty-intelligence-mcp

27,681 accepted findings from Sherlock and Code4rena, structured as searchable MCP tools:

list_vulnerability_patterns (free) - given a protocol type (LENDING, DEX, BRIDGE, TOKEN), returns historical acceptance rates per vulnerability category. Useful for prioritizing what Plamen's agent should look for first.

scan_contract (5 USDC on Base via x402) - submits a GitHub repo for Al-Mizaan v3 analysis, an independent 7-gate framework. Could serve as a second opinion or pre-scan before a deeper Plamen audit.

get_scan_report (free) - polls and returns report URL.

Config for Claude Code:

{"mcpServers":{"bug-bounty-intelligence":{"command":"npx","args":["-y","github:holistis/bug-bounty-intelligence-mcp"]}}}

The possible integration: Plamen's agent could call list_vulnerability_patterns early in an audit to get a data-driven target list, then focus its deeper analysis on the historically high-acceptance-rate categories for that protocol type.

Happy to share more on the framework or data if useful.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions