Skip to content

Security: lamentierschweinchen/kommit

Security

SECURITY.md

Security policy

Kommit is a hackathon/devnet-grade prototype submitted to Solana Frontier (May 2026). It is not third-party audited, not production-ready, and currently only deployed on Solana devnet. Mainnet is whenever-if-ever — the architecture is mainnet-ready, but the button has not been pressed.

If you find a security issue, please follow this disclosure path before opening a public issue or PR.

Disclosure

Email: lukas.c.seel@gmail.com

Please put [Kommit security] in the subject line. We aim to acknowledge within 72 hours and provide a fix or mitigation timeline within 7 days for in-scope issues. Out-of-scope issues are acknowledged but not fixed.

We follow coordinated vulnerability disclosure:

  • Don't publicly disclose before we've had a reasonable opportunity to mitigate.
  • We will credit you in the fix commit + release notes if you want.
  • We won't pursue legal action against good-faith research that respects the boundaries below.

In scope

  • The Anchor program at programs/kommit/ (devnet program ID GxM3sxMp4FyrkHK4g1DaDrmwYLrwd2BJKxqKZqvGgkc3)
  • The off-chain webhook indexer at web/src/app/api/webhook/helius/
  • The Supabase schema migrations at migrations/supabase/
  • The Next.js frontend at web/
  • The mainnet deploy scripts at scripts/
  • This repository's secrets posture (any reachable production key, private repo data, etc.)

Severity rubric

  • Critical: principal-loss path, unauthorized fund movement, full admin compromise, RCE on production infra, secret exfiltration enabling any of the above.
  • High: unauthorized state writes, RLS bypass that exposes user identity, supply-chain compromise, denial-of-service that costs the platform > $1k/day.
  • Medium: information disclosure of non-PII, privilege escalation between user roles, persistent XSS, CSRF on state-changing routes.
  • Low: missing security headers, weak password policies on admin tooling, non-exploitable code-quality issues with security flavor.
  • Info: best-practice deviations, architectural notes.

Out of scope

  • Reports against web-legacy-anchor-ref/ — this directory exists as a code reference for an in-progress migration and is not deployed anywhere. Issues there are not security-relevant.
  • Theoretical vulnerabilities without a working exploit on devnet.
  • Social engineering against Kommit team members.
  • Physical attacks against Lukas's hardware.
  • Reports about third-party services (Privy, Supabase, Helius, Kamino, Solana itself) that do not affect Kommit's specific integration. Please report those upstream.
  • Self-XSS, missing rate limits on demo endpoints, or other low-impact theoretical issues that require unrealistic preconditions.

Known limitations

These are documented and intentional, not bugs:

  • No KYC — private beta is whitelisted; no KYC plumbing in v0.5. KYC plumbing arrives with v1 fiat rails (card / SEPA partners enforce it upstream).
  • Curation is centralized — admin-curated invite-only in v0.5; less-centralized paths (DAO, staked-reputation) explored in v2+.
  • Single yield-source adapter in v0.5 (Kamino USDC reserve). Second adapter (marginfi or Jupiter Lend) is a v1 task.
  • Demo controls (?as= persona override + floating <DemoControls />) are gated behind NODE_ENV !== "production" for production builds. Mock auth never runs in production.
  • SECURITY_REVIEW.md — internal self-audit of the Anchor program with file:line citations. Verdict: hackathon-private-beta-grade. Pre-scaling upgrade items (multisig admin + multisig upgrade authority + third-party audit) called out explicitly.
  • SECURITY_HARDENING.md (at workspace parent) — full-stack security review by an internal Codex pass. Verdict: HOLD-WITH-FIXES at time of writing; tracked findings burned down in this commit cycle.

Structural risks (not bugs — architectural tradeoffs)

For the risk surfaces that exist by virtue of the architecture itself — smart-contract risk on Kommit's program, yield-source risk on Kamino, oracle risk, counterparty risk, regulatory framing, operational/curation risk — see RISK.md. Bug reports go through this doc; structural-tradeoff conversations belong there.

There aren't any published security advisories