Skip to content

RFC: Cryptographic evidence chain for PentAGI pentest operations #235

@tomjwxf

Description

@tomjwxf

Problem

PentAGI runs automated penetration testing with AI agents. Security assessments produce findings that need to be demonstrably authentic and tamper-proof — a modified pentest report could hide vulnerabilities or fabricate findings. Currently, there is no cryptographic evidence chain proving what tools were run, what was discovered, and that the results are unmodified.

Proposal

Integrate Ed25519 receipt signing into PentAGI's tool execution pipeline. Each pentest tool invocation would produce a signed receipt creating a verifiable evidence chain:

Recon (receipt_001) → Port Scan (receipt_002) → Exploit Attempt (receipt_003) → Report (receipt_004)

This transforms pentest results from "trust me" reports into cryptographically verifiable evidence packages.

Reference

protect-mcp (MIT, v0.5.3). Receipt format: IETF Internet-Draft. The receipt chain/DAG visualization is built-in via npx protect-mcp trace <id>.

Happy to discuss and contribute.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions