Skip to content

Security: source-origin/l5-protocol

Security

SECURITY.md

Security Policy

ORIGIN L5 is a settlement layer for the AI-agent economy. Settlement code is a high-trust surface: a bug here can move or strand value. We take reports seriously and we would rather hear from you first.

⚠️ Status: research-grade. These contracts are not yet professionally audited. Do not use with real funds. Our own dated findings and fixes are published in docs/SELF-AUDIT-2026-09-25.md.

Reporting a vulnerability

Please do not open a public issue for a security bug.

  • Private channel (preferred): use GitHub's private vulnerability reporting — repo Security tab → Report a vulnerability.
  • If you cannot use that, open a minimal public Issue that says only "security report, please contact me privately" with no technical detail, and a maintainer will reach out.

Please include: affected contract + function, a description of the impact, and — if you have one — a failing test or a Foundry PoC.

What we consider in scope

  • Fund loss / theft, double-spend, or stranding of escrowed or channel value.
  • Authorization bypass — reaching a privileged path (onlyOwner / onlyAdmin / party checks) without being authorized.
  • Replay of a signature or a receipt across receipts, chains, or deployments.
  • Unbounded authorization — a delegated spend that exceeds, or escapes, its on-chain policy cap.
  • State corruption that leaves a receipt/channel/escrow in an unrecoverable or dishonest state.

Out of scope: gas optimization, style, and known open design items already disclosed in the self-audit record (§6 there: multisig/timelock, delegation debit semantics, YUAN mint centralization). If you disagree with a disclosed design decision, we want that discussion too — open a Discussion.

Our commitments

  • Acknowledge a report within 48 hours.
  • Give you a first assessment (in-scope? severity?) within 5 business days.
  • Credit you in the fix and in this file, unless you ask to stay anonymous.
  • Coordinate disclosure — we will not publish details before a fix ships, and we will tell you when it does.

Hard rules for our own code (from Constitution L0)

  • No change may weaken withdrawal / revoke authority — humans stay in control.
  • Any path that could double-spend or lose funds is a top-priority issue, even mid-discussion.
  • Settlement is final; reversibility belongs to the authorization and custody layers, never to rewriting a settled receipt (see self-audit §5).

量子总督 (Quantum Governor) 👽 · for the ORIGIN L5 maintainers

There aren't any published security advisories