momentic@3.48.0
 [![npm]](https://www.npmjs.com/package/momentic)
AI Security Review
scanned 3mo ago · by lpm-firewall-aiThe package has an explicit user-command path that writes AI-agent skill files into project control surfaces, warranting a warning under policy. Direct inspection found no lifecycle script or confirmed malicious credential-exfiltration chain.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
User runs `momentic install-skills` with an editor choice.
Impact
Modifies the selected project agent skill directory with bundled package content.
Mechanism
Explicit project AI-agent skill installation
AI rationale
The package has an explicit user-command path that writes AI-agent skill files into project control surfaces, warranting a warning under policy. Direct inspection found no lifecycle script or confirmed malicious credential-exfiltration chain.
Evidence
package.jsonbin/cli.jsskills/momentic-test.mdskills/*.md.cursor/skills/*/SKILL.md.claude/skills/*/SKILL.md.agents/skills/*/SKILL.md.opencode/skills/*/SKILL.md.github/skills/*/SKILL.md
Decision evidence
public snapshotAI called this Suspicious at 95.0% confidence as Dangerous Capability with low false-positive risk.
Evidence for warning
- Explicit `install-skills` CLI command writes bundled skills into Cursor, Claude, Codex, OpenCode, or Copilot project directories.
- The package provides a manifest-reachable CLI, so this mutation is user-triggerable.
Evidence against
- package.json has no lifecycle scripts, so installation does not execute the CLI.
- Agent-directory mutation is limited to the explicit deprecated `install-skills` command.
- Source rejects reads of common credential and secret file paths.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShellWebSocket
HighEntropyStringsMinifiedPossibleObfuscationTelemetryUrlStrings
NoLicense
Source & flagged code
17 flaggedpackage.jsonView file
•name = momentic; similarTo = moment
Medium
Typosquat Name
Package name is suspiciously similar to a popular package name.
package.jsonView on unpkgbin/cli.jsView file
11import oJ from 'truncate-json';
L12: import Fl, { supportsColorStderr, supportsColor, Chalk } from 'chalk';
L13: import GOe from 'safe-stable-stringify';
...
L30: import YF, { randomUUID, createHash, createHmac, timingSafeEqual } from 'crypto';
L31: import * as dnt from 'child_process';
L32: import { execFile, execSync, exec, execFileSync, spawn as spawn$1 } from 'child_process';
...
L37: import { registerInstrumentations } from '@opentelemetry/instrumentation';
L38: import { UndiciInstrumentation } from '@opentelemetry/instrumentation-undici';
L39: import { resourceFromAttributes } from '@opentelemetry/resources';
...
L160: Use this only after the available tools cannot establish a defensible root cause. Do not use OTHER while targeted current-run step results, container screenshots,
Critical
Credential Exfiltration
Source appears to send environment or credential material to an external endpoint.
bin/cli.jsView on unpkg · L11Findings
4 Critical11 High4 Medium6 Low
CriticalCredential Exfiltrationbin/cli.js
CriticalBrowser Credential Phishingbin/cli.js
CriticalTrigger Reachable Dangerous Capabilitybin/cli.js
CriticalPrevious Version Dangerous Deltabin/cli.js
HighChild Processbin/cli.js
HighShellstatic/assets/index-yn3LtkYW.js
HighSame File Env Network Executionbin/cli.js
HighCredential Redirect Persistencebin/cli.js
HighEntrypoint Foreign Package Code Overwritebin/cli.js
HighCommand Output Exfiltrationbin/cli.js
HighRemote Agent Bridgebin/cli.js
HighTrigger Reachable Credential Exfiltration
Affected versions and remediation
This report applies to momentic@3.48.0.
See all versions of momentic for other recorded verdicts.
- Review the evidence and your use of momentic@3.48.0 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.