Skip to content

Security: mertksk/atlas-agent

Security

.github/SECURITY.md

Security Policy

Atlas Agent is a non-custodial application: every fund-moving action (the usage fee, allocations, cspr.trade swaps) is signed by the user's own Casper Wallet. The server never holds a user's private key and cannot move a user's funds. It runs on Casper Testnet with test-only tokens.

Reporting a vulnerability

If you find a security issue, please do not open a public issue. Instead:

  • Use GitHub's Report a vulnerability (Security → Advisories) to open a private report, or
  • email the maintainer at the address on the GitHub profile.

Please include: a description, reproduction steps, affected component (agent/, web/, services/, or contracts/), and impact. We aim to acknowledge within 72 hours.

Scope & trust model

  • In scope: the agent API (agent/), the dashboard (web/), the x402 data services (services/), and the Odra contracts (contracts/).
  • Funds safety: all transfers require the connected wallet's signature — a server compromise cannot move a user's CSPR.
  • Known accepted limitation: per-wallet sessions are keyed by the (public) wallet address with no proof-of-control on reads, so someone who knows a public key can view that session's off-chain pending recommendations (never move funds). A production deployment would add signed-challenge session auth.

Handling of secrets

Secrets (API tokens, RPC keys, the agent's operating key) live only in the server .env and are never committed. Secret scanning + push protection are enabled on this repository.

There aren't any published security advisories