eve@0.79.2
Filesystem-first framework for durable backend AI agents that run anywhere.
AI Security Review
scanned 8d ago · by lpm-firewall-aiSource inspection found no install lifecycle hook or concrete attack behavior. The flagged Vercel, CLI, MCP skills, and HTTP Basic code supports ordinary package functionality.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review
Trigger
Vercel authentication occurs when the application requests Vercel model access; CLI build steps occur when the CLI needs rebuilding.
Impact
No harmful effect was established from these behaviors.
Mechanism
The inspected code restricts OAuth endpoints to Vercel, checks access with Vercel’s AI Gateway, and invokes local build tools using Node.
AI rationale
Source inspection found no install lifecycle hook or concrete attack behavior. The flagged Vercel, CLI, MCP skills, and HTTP Basic code supports ordinary package functionality.
Decision evidence
public snapshotAI called this Clean at 94.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The package manifest’s scripts are build, check, and test tasks; no install lifecycle hook is present.
- The CLI runs local build tools through Node when it needs to build the CLI.
- Vercel OAuth endpoints are restricted to vercel.com and api.vercel.com; the authorization check sends a bearer token to Vercel’s AI Gateway.
- The MCP skills code serves files through its provided file reader.
- The HTTP Basic auth code compares supplied credentials with the configured strategy credentials.
Evidence against
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNativeBindingsNetworkShell
HighEntropyStringsMinifiedPossibleObfuscationUrlStrings
Source & flagged code
8 flagged · loading sourcebin/eve.jsView file
2L3: import { spawn } from "node:child_process";
L4: import { access, readdir, realpath, stat } from "node:fs/promises";
High
2Cross-file remote execution chain: bin/eve.js spawns dist/src/compiled/@ai-sdk/anthropic/index.js; helper contains network access plus dynamic code execution.
L2:
L3: import { spawn } from "node:child_process";
L4: import { access, readdir, realpath, stat } from "node:fs/promises";
...
L9: const require = createRequire(import.meta.url);
L10: const packageJson = require("../package.json");
L11: const packageNodeEngine = packageJson.engines?.node;
...
L331:
L332: await new Promise((resolveFlush) => stream.write("", () => resolveFlush()));
L333: }
...
L365: }
L366: process.exitCode = 1;
L367: } finally {
High
Cross File Remote Execution Context
Source spawns a local helper that also contains network and dynamic execution context; review data flow before blocking.
bin/eve.jsView on unpkg · L28L9: const require = createRequire(import.meta.url);
L10: const packageJson = require("../package.json");
Medium
Dynamic Require
Package source references dynamic require/import behavior.
bin/eve.jsFindings
1 Critical5 High5 Medium6 Low
CriticalHardcoded Runtime Data Exfiltrationdist/src/internal/model-auth/vercel.js
HighChild Processbin/eve.js
HighShell
HighCross File Remote Execution Contextbin/eve.js
HighSemantic Analysis Limiteddist/src/internal/mcp/skills.js
HighKnown Malware Source Similaritydist/src/channel/auth/http-basic.js
MediumDynamic Requirebin/eve.js
MediumNetwork
MediumEnvironment Vars
MediumShips Build Helperdist/src/compiled/just-secrets/windows.ps1
MediumStructural Risk Force Deep Review
LowScripts Present
LowEval
Low
Affected versions and remediation
This report applies to eve@0.79.2.
See all versions of eve for other recorded verdicts.
- Review the evidence and your use of eve@0.79.2 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.