Skip to content

Security: leopardracer/FLETCH

Security

SECURITY.md

Security Policy

Scope

FLETCH is a read-only intelligence layer: it reads Robinhood Chain state through RPC/Blockscout and never holds funds, signs transactions, or exposes a private key or custodial path. There is no execution path in this repository (see the README). That materially limits the blast radius of a bug here, but three classes of issue still matter and are in scope:

  • A crafted RPC response, event log, or API input causing incorrect risk/ score output that could mislead a user into a bad trade
  • A way to make the API (src/api/server.ts) crash, leak another caller's data, or execute arbitrary code
  • Exposure of .env secrets (BLOCKSCOUT_API_KEY, RPC_URL) through logs, error responses, or the dashboard

Out of scope: issues that require an already-compromised RPC endpoint, denial-of-service that just needs raw request volume, or "the score disagrees with my opinion of a token" (open a normal issue for that instead).

Reporting a Vulnerability

Please do not open a public GitHub issue for a security report.

Use GitHub Security Advisories for this repository — it's private between you and the maintainer until a fix ships. Include:

  • The affected file/endpoint and, ideally, a minimal repro
  • What real-world impact you'd expect (misleading score? crash? secret exposure?)
  • Whether it needs a live RPC connection to trigger, or reproduces against the deterministic test suite

You should get an initial response within 5 business days. If a report is confirmed, a fix is prioritized ahead of feature work and credited in the advisory (unless you'd rather stay anonymous) once it ships.

Supported Versions

FLETCH is pre-1.0 (0.x). Only the main branch is supported — there are no maintained release branches to backport a fix to yet.

There aren't any published security advisories