create-mastra@1.32.1
Create Mastra apps with one command
AI Security Review
scanned 22d ago · by lpm-firewall-aiThe configurable authentication destination creates conditional credential exposure, while explicit agent skill installation adds a guarded agent capability. Neither behavior establishes malicious intent or qualifies for publication blocking.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
Run create-mastra and accept observability setup while MASTRA_PLATFORM_API_URL points to another recipient; skill installation runs unless disabled.
Impact
Conditional exposure of existing Mastra session credentials and installation of agent instructions in the generated project.
Mechanism
Authentication sends saved access and refresh tokens to the configurable API base without a destination allowlist. Project setup invokes npx to install mastra-ai/skills for detected agents.
AI rationale
The configurable authentication destination creates conditional credential exposure, while explicit agent skill installation adds a guarded agent capability. Neither behavior establishes malicious intent or qualifies for publication blocking.
Evidence
dist/index.jsdist/credentials-6AY2wEWQ-DCCCg72E.jsdist/client-BzuP6FOB-C6hwDwJW.js.env
Network endpoints1
platform.mastra.ai
Decision evidence
public snapshotAI called this Suspicious at 96.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The platform API destination accepts an unrestricted MASTRA_PLATFORM_API_URL override.
- Authentication loads saved Mastra credentials and sends the access token for verification and the refresh token for renewal to that destination.
- The active create command invokes authentication after the user accepts observability setup; this consent does not restrict the credential recipient.
- Project setup installs Mastra skills for detected coding agents through npx with automatic confirmation; --no-skills disables this behavior.
- The flagged MCP configuration location only constructs paths and performs no configuration write.
Evidence against
- package.json has no automatic installation lifecycle hooks or runtime self-dependency.
- The default authentication destination is Mastra's platform, and observability requires an interactive choice.
- No covert recipient, destructive behavior, or malicious intent was established.
Behavioral surface
ChildProcessEnvironmentVarsFilesystemNetwork
Source & flagged code
8 flagged · loading sourcedist/index.jsView file
5import { EventEmitter, addAbortListener, on, once, setMaxListeners } from "node:events";
L6: import { ChildProcess, execFile, spawn, spawnSync } from "node:child_process";
L7: import path, { basename, delimiter, dirname, isAbsolute, normalize, resolve } from "node:path";
...
L35: * Constructs the CommanderError class
L36: * @param {number} exitCode suggested exit code which could be used with process.exit
L37: * @param {string} code an id string representing the error
...
L181: * @return {string}
L182: * @private
L183: */
...
L196: * TypeScript import types for JSDoc, used by Visual Studio Code IntelliSense and `npm run typescript-checkJS`
L197: * https://www.typescriptlang.org/docs/handbook/jsdoc-supported-types.html#import-types
L198: * @typedef { import("./argument.js").Argument } A
Critical
Hardcoded Runtime Data Exfiltration
Source sends credentials or rich application records to a package-controlled external receiver enabled by default.
dist/index.jsView on unpkg · L55Trigger-reachable chain: manifest.main -> dist/index.js
L5: import { EventEmitter, addAbortListener, on, once, setMaxListeners } from "node:events";
L6: import { ChildProcess, execFile, spawn, spawnSync } from "node:child_process";
L7: import path, { basename, delimiter, dirname, isAbsolute, normalize, resolve } from "node:path";
...
L35: * Constructs the CommanderError class
L36: * @param {number} exitCode suggested exit code which could be used with process.exit
L37: * @param {string} code an id string representing the error
...
L181: * @return {string}
L182: * @private
L183: */
...
L196: * TypeScript import types for JSDoc, used by Visual Studio Code IntelliSense and `npm run typescript-checkJS`
L197: * https://www.typescriptlang.org/docs/handbook/jsdoc-supported-types.html#import-types
Findings
2 Critical6 High3 Medium5 Low
CriticalHardcoded Runtime Data Exfiltrationdist/index.js
CriticalTrigger Reachable Dangerous Capabilitydist/index.js
HighChild Processdist/index.js
HighSame File Env Network Executiondist/index.js
HighCommand Output Exfiltrationdist/credentials-6AY2wEWQ-DCCCg72E.js#virtual:normalized:round1
HighSandbox Evasion Gated Capabilitydist/index.js
HighTrigger Reachable Command Output Exfiltrationdist/credentials-6AY2wEWQ-DCCCg72E.js#virtual:normalized:round1
HighSemantic Analysis Limiteddist/credentials-6AY2wEWQ-DCCCg72E.js#virtual:normalized:round1
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
Affected versions and remediation
This report applies to create-mastra@1.32.1.
See all versions of create-mastra for other recorded verdicts.
- Review the evidence and your use of create-mastra@1.32.1 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.