pintumcp@1.2.8
Agent Alerts MCP — desktop notifications with sound for AI coding agents
AI Security Review
scanned 14d ago · by lpm-firewall-aiThe postinstall automatically writes agent configuration and installs plugins or hooks across detected tools. This is an unconsented lifecycle mutation of AI-agent control surfaces, so the package should be blocked.
Static reason
One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source
Trigger
Installing the package through npm runs its postinstall hook.
Impact
Agent tools may load the package’s server, plugin, or hooks after restart without a separate setup command.
Mechanism
The hook launches the CLI install command, which writes MCP configuration entries and adds OpenCode and Claude Code integrations when their directories are present.
Policy narrative
The npm postinstall launches the install command, which discovers installed agent tools and writes integrations into their configuration surfaces. It also installs an OpenCode plugin and Claude Code command hooks automatically when their directories exist. These changes occur as part of npm installation, without a separate user-invoked setup command.
AI rationale
The postinstall automatically writes agent configuration and installs plugins or hooks across detected tools. This is an unconsented lifecycle mutation of AI-agent control surfaces, so the package should be blocked.
Evidence
package.jsoninstall.jsbin/pintumcp.jslib/configure.jslib/hooks.js
Decision evidence
public snapshotAI identified this as a dangerous AI-agent capability at 97.0% confidence with low false-positive risk.
Evidence for policy risk
- package.json defines a postinstall hook that runs install.js, which launches the CLI install command.
- The install command detects AI tools and configures each detected tool automatically.
- During installation, the CLI copies a plugin into OpenCode’s plugins directory and adds Claude Code hooks when its settings directory exists.
- The hook integration adds command hooks for multiple Claude Code events and saves them to the settings file.
- MCP configuration writes the package’s server entry into detected agent configuration files.
Evidence against
Behavioral surface
ChildProcessEnvironmentVarsFilesystemShell
HighEntropyStrings
Source & flagged code
7 flagged · loading sourcepackage.jsonView file
•scripts.postinstall = node install.js
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg9"scripts": {
"postinstall": "node install.js"
},
"keywords":
High
Suspicious Lifecycle Evidence
package.json defines a postinstall hook that runs install.js, which launches the CLI install command.
package.jsonView on unpkg · L9•scripts.postinstall = node install.js
Medium
Ambiguous Install Lifecycle Script
Install-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgpython/tray_cross.pyView file
Findings
5 High4 Medium3 Low
HighInstall Time Lifecycle Scriptspackage.json
HighKnown Malware Source Similaritybin/pintumcp.js
HighPrevious Version Dangerous Deltabin/pintumcp.js
HighSuspicious Lifecycle Evidencepackage.json
HighSuspicious Lifecycle Evidenceinstall.js
MediumAmbiguous Install Lifecycle Scriptpackage.json
MediumEnvironment Vars
MediumShips Build Helperpython/tray_cross.py
MediumStructural Risk Force Deep Review
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
Affected versions and remediation
This report applies to pintumcp@1.2.8.
See all versions of pintumcp for other recorded verdicts.
- Avoid installing pintumcp@1.2.8. Remove it from direct dependencies and check your lockfile for transitive copies.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
- If this version ran, investigate the affected machine and build environment. Rotate credentials it could access and rebuild from a trusted environment.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.