Skip to content

Security: whereiskurt/klanker-maker

Security

SECURITY.md

Security Policy

Klanker Maker is a personal project (see NOTICE.md). There is no commercial support, no SLA, and no formal security team. The notes below describe how to report vulnerabilities and what to expect in return.

Reporting a Vulnerability

Please do not open a public GitHub issue for security-sensitive reports.

Email whereiskurt@gmail.com with [klanker-maker-security] in the subject and include:

  • A description of the vulnerability
  • Steps to reproduce (proof-of-concept code or commands welcome)
  • Affected versions, commits, or default profiles
  • Your assessment of impact and any suggested mitigations
  • Whether you'd like to be credited in any eventual release notes

I will acknowledge receipt within 7 days and aim to provide a more detailed response within 30 days indicating next steps and a rough timeline. As a side project this may slip; please feel free to email a polite nudge if you haven't heard back.

If you'd like to encrypt the report, ask in your first email and I will respond with a public key.

Scope

In scope:

  • The km CLI and any binaries built from this repository
  • Sidecar binaries (DNS proxy, HTTP proxy, audit log, tracing, km-slack)
  • Lambda function source under cmd/*-handler/, cmd/*-refresher/, and the Slack bridge
  • Terraform / Terragrunt modules in infra/
  • The eBPF programs in pkg/ebpf/
  • The default profiles shipped in profiles/

Out of scope:

  • Vulnerabilities in third-party services Klanker Maker integrates with (AWS, Anthropic, Slack, GitHub, OpenAI). Please report those to the respective vendors.
  • Issues that require operator misconfiguration to exploit. The threat model assumes the operator (the human running km) is trusted; the agents running inside sandboxes are not.
  • Theoretical attacks against forks where the fork has materially changed the default security model (e.g., disabled the SCP, removed the eBPF enforcer, bypassed the proxy).
  • Bugs in unmerged branches.

The defense-in-depth model is documented in README.md § Security Model and docs/security-model.md. Reports demonstrating gaps in the documented controls are most useful.

Disclosure

This is a personal project; I will work with reporters in good faith on coordinated disclosure but cannot commit to a fixed embargo or guarantee any particular timeline. As a rough guide:

  • Critical issues with active exploit potential: I'll prioritize a fix before public disclosure where possible.
  • Lower-severity issues: I may publish the fix and a brief advisory together.

If you intend to publish independently after a reasonable period (e.g., 90 days from acknowledgement), please tell me up front so I can plan around it.

No Bug Bounty

There is no monetary bug bounty. Reports are accepted out of goodwill, and credit is given in release notes if the reporter would like to be named.

Supported Versions

Only the main branch and the most recent tagged release receive security updates. Older versions are unsupported.

See Also

  • LICENSE - the warranty disclaimer that applies to this software.
  • NOTICE.md - personal-project status, no employer affiliation.
  • CONTRIBUTING.md - DCO and contributor warranty terms.

There aren't any published security advisories