eve@0.30.8
Filesystem-first framework for durable backend AI agents that run anywhere.
AI Security Review
scanned 3mo ago · by lpm-firewall-aiThe flagged primitives belong to explicit framework/CLI and Vercel integration paths. Source inspection found no lifecycle hook, stealth persistence, broad agent-control mutation, or concrete exfiltration chain.
Decision evidence
public snapshot- Vendored Vercel OIDC helper sends hostname/platform in a Vercel OAuth user-agent.
- Sandbox backend can run commands and manipulate files only when application code uses it.
- package.json has no preinstall/install/postinstall/prepare hook.
- bin/eve.js executes only when the user invokes the eve CLI.
- bin/eve.js child processes are local build/TypeScript bootstrap commands.
- start-production-server.js starts the user's built local server and polls its local health route.
- token-util-DURSL1l8.js uses Vercel OAuth/project-token endpoints, not an unrelated collector.
- No foreign AI-agent configuration mutation or unsolicited credential/file exfiltration found.
Source & flagged code
11 flagged · loading sourcePackage source references dynamic require/import behavior.
bin/eve.jsView on unpkg · L8Package source references a known benign dynamic code generation pattern.
dist/src/compiled/gray-matter/index.jsView on unpkg · L43Findings
Affected versions and remediation
This report applies to eve@0.30.8.
See all versions of eve for other recorded verdicts.
- Review the evidence and your use of eve@0.30.8 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.