byreal-hermes-deploy-native
Warn
Audited by Snyk on May 20, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.90). The skill's required install flow explicitly downloads and installs code/assets from public GitHub (Step 2's curl of the uv release and Step 3's git clone of HERMES_AGENT_REPO) — untrusted third‑party content that is fetched and executed/installed as part of the workflow and thus can materially influence agent behavior.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.90). The install flow fetches and executes remote artifacts at runtime — notably the uv release tarball from https://github.com/astral-sh/uv/releases/download/${UV_VERSION}/${UV_ASSET} and the hermes-agent git repo at https://github.com/NousResearch/hermes-agent.git — which are required dependencies and result in executing remote code on install.
MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).
- Direct money access detected (high risk: 1.00). The skill explicitly handles blockchain wallets and prepares the agent for on-chain operations. It runs wallet-info via an agent-token script and byreal-cli to obtain wallet addresses, writes "USER.md" with those addresses, and states "RealClaw Privy server-side wallets — Hermes uses these for on-chain ops." It also notes TOOLS.md is needed for chain addresses/program IDs. These are specific crypto/wallet integrations (not generic tooling), so the skill enables crypto on-chain operations and therefore qualifies as direct financial execution capability.
MEDIUM W013: Attempt to modify system services in skill instructions.
- Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill instructs the agent to modify user shell files, create/symlink and delete directories, write configs with secrets, kill and restart processes, clone and install software and even conditionally run sudo apt-get to install system packages—actions that change the host state and can require elevated privileges.
Issues (4)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W012
MEDIUMUnverifiable external dependency detected (runtime URL that controls agent).
W009
MEDIUMDirect money access capability detected (payment gateways, crypto, banking).
W013
MEDIUMAttempt to modify system services in skill instructions.
Audit Metadata