registry / @mnemonik/cli / 7.173.0
@mnemonik/cli@7.173.0
Mnemonik installer and local project helper
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe manifest has no install hooks, and the flagged eval is used by bundled Tree-sitter WebAssembly bindings. The reviewed outbound request carries project hashes to the package's own service; no malicious credential flow or unrelated execution was found.
Static reason
One or more suspicious static signals were detected.
Trigger
A user invokes the CLI or a Mnemonik hook reaches project-context binding.
Impact
No credential theft, concealed remote execution, or unrelated account action was established.
Mechanism
The helper gathers Git remote information, derives project hashes, and posts context to the configured server. Tree-sitter's bundled WebAssembly glue uses eval to create module bindings.
AI rationale
The manifest has no install hooks, and the flagged eval is used by bundled Tree-sitter WebAssembly bindings. The reviewed outbound request carries project hashes to the package's own service; no malicious credential flow or unrelated execution was found.
Evidence
The runtime writes a context cache at input.stateFile; its resolved path is not fixed in the inspected source.
Network endpoints1
api.mnemonik.dev (default; configurable through MNEMONIK_API_RESOURCE)
Decision evidence
public snapshotAI called this Clean at 88.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The manifest describes an installer and local project helper, has no lifecycle scripts, and exposes a CLI plus library entry points.
- The flagged eval constructs JavaScript bindings from strings in the Tree-sitter WebAssembly module.
- The runtime POST sends project context to the configured server; the body includes HMAC and repository fingerprint hashes.
- The default API origin is the package's own api.mnemonik.dev service.
Evidence against
- The CLI runs its bootstrap only when invoked; the manifest defines no install-time hook.
- The git-scm.com matches are documentation links in scanner code, not credential destinations.
- The reviewed request body contains project identifiers and hashes, with no raw environment values or credential values shown.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStrings
Source & flagged code
8 flaggeddist/diagnostics.jsView file
923import { createHash, createPublicKey, randomUUID, timingSafeEqual, verify as verifySignature } from "node:crypto";
L924: import { execFile, execFileSync } from "node:child_process";
L925: function windowsCommand(file) {
High
Child Process
Package source references child process execution.
dist/diagnostics.jsView on unpkg · L9235256var func = `(${args}) => { ${body} };`;
L5257: ASM_CONSTS[start] = eval(func);
L5258: }
High
1import { createRequire as __createRequire } from "node:module"; const require = __createRequire(import.meta.url);
L2: var __create = Object.create;
Medium
Dynamic Require
Package source references dynamic require/import behavior.
dist/diagnostics.jsView on unpkg · L1Findings
1 Critical6 High4 Medium4 Low
CriticalRemote Response Code Executiondist/scanner/service.js
HighChild Processdist/diagnostics.js
HighEvaldist/diagnostics.js
HighSame File Env Network Executiondist/install/journey.js
HighCommand Output Exfiltrationdist/runtime/signers.js
HighSandbox Evasion Gated Capabilitydist/install/journey.js
HighCross File Remote Execution Contextdist/auth/credentials.js
MediumDynamic Requiredist/diagnostics.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowFilesystem
LowHigh Entropy Strings
Affected versions and remediation
This report applies to @mnemonik/cli@7.173.0.
See all versions of @mnemonik/cli for other recorded verdicts.
- Review the evidence and your use of @mnemonik/cli@7.173.0 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.