Skip to content

Security: poyzfi/poyz

Security

SECURITY.md

Security Policy

Reporting a vulnerability

Report privately. Do not open a public issue, and do not describe the finding in a pull request.

Use GitHub's private vulnerability reporting on this repository: Security -> Report a vulnerability (https://github.com/poyzfi/poyz/security/advisories/new). It creates a private advisory thread visible only to the maintainers and to you.

If that route is unavailable, send a direct message to @poyzfi asking for a private channel. Do not put finding details in a public post.

Include what you have: the affected instruction or module, the conditions required, and what an attacker gains. A proof-of-concept against a local validator is useful but not required to report.

What to expect

Stage Target
Acknowledgement of the report 72 hours
Initial assessment and severity 7 days
Fix or documented mitigation for a confirmed high-severity finding 30 days
Public disclosure after a fix ships, coordinated with the reporter

If a report goes unacknowledged past 72 hours, ping the same channel. Silence is a failure on our side, not a rejection.

Scope

In scope:

  • The poyz Anchor program under programs/ -- account validation, arithmetic, authority and upgrade paths, oracle handling, keeper bond and slash accounting, and the rebalance proof chain.
  • The generated idl/poyz.json where it misdescribes an instruction in a way that would lead a client to build the wrong transaction.
  • Documentation in docs/ where an incorrect specification would lead an integrator into an unsafe integration.

The TypeScript SDK and the command line interface are covered by the security policy of poyzfi/poyz-sdk, not by this one.

Out of scope:

  • Findings that require a compromised signer key or a maliciously modified client.
  • Denial of service against public RPC endpoints, or third-party infrastructure that this project does not operate.
  • Market risk. Negative funding, venue outages and liquidation are inherent to the design and are documented in docs/risk-spec.md. Those are disclosures, not vulnerabilities. A defect in how the protocol handles one of them is in scope.
  • Automated scanner output with no demonstrated impact.

Deployment status

The program is not deployed to Solana mainnet. Anchor.toml targets localnet and declare_id! currently holds the Anchor placeholder id Fg6PaFpoGXkYsidMpWTK6W2BeZ7FEfcYkg476zPFsLnS. There is no live program to attack yet, and no funds are at risk today. Reports against the source are still welcome and are the cheapest time to fix anything.

This project has not been audited. docs/security.md sets out the threat model, the authority and upgrade structure, and the oracle safeguards the program relies on. Read it before integrating.

Safe harbour

Good-faith research on a local validator or a private fork is welcome, and we will not pursue action over it. Do not test against third-party mainnet infrastructure, do not access data that is not yours, and do not run anything that degrades service for others.

There aren't any published security advisories