eve@0.64.1
Filesystem-first framework for durable backend AI agents that run anywhere.
AI Security Review
scanned 1mo ago · by lpm-firewall-aieve 0.64.1 is an agent framework whose inspected entrypoints run only when a user launches the CLI or calls auth helpers. Vercel tokens stay on Vercel origins, and there is no install lifecycle hook or foreign agent-control write.
Decision evidence
public snapshot- package.json identifies eve 0.64.1 as a published package and does not declare preinstall, install, or postinstall hooks.
- bin/eve.js is the user-launched CLI and resolves the packaged dist CLI entrypoint.
- Vercel model auth accepts OAuth endpoints only for https://vercel.com or https://api.vercel.com and sends the session token to Vercel AI Gateway.
- Mock model fixtures only build local test replies and tool inputs.
- HTTP basic channel auth compares a caller-supplied username and password locally with a timing-safe hash.
- Searches of package.json, bin, and dist/src found no preinstall, install, or postinstall scripts.
- Credential refresh is gated by an origin allowlist and is aimed at the Vercel login service, not an unrelated recipient.
Source & flagged code
10 flagged · loading sourceSource spawns a local helper that also contains network and dynamic execution context; review data flow before blocking.
bin/eve.jsView on unpkg · L2Source file is highly similar to a previously finalized malicious package; route for source-aware review.
Findings
Affected versions and remediation
This report applies to eve@0.64.1.
See all versions of eve for other recorded verdicts.
- Review the evidence and your use of eve@0.64.1 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.