registry / @funny-booth/agent-core / 0.1.26
@funny-booth/agent-core@0.1.26
古い版です。最新版を使ってください
AIエージェント起動CLI（共有ランチャー + MCPサーバー）
AI Security Review
scanned 9d ago · by lpm-firewall-aiThe launcher can install and execute portal-configured MCP extensions with access to the portal token and selected local secrets, so this warrants a warning. The behavior is runtime extension setup and no concrete malicious action or unconsented install-time mutation was established.
Static reason
One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source
Trigger
A user runs one of the package's CLI launchers.
Impact
A malicious or compromised extension could act with the credentials provided to its process; source inspection does not show that these extensions exfiltrate them.
Mechanism
The launcher fetches MCP metadata from the configured portal, prepares package executables, and supplies them through the agent's MCP configuration with selected environment values.
AI rationale
The launcher can install and execute portal-configured MCP extensions with access to the portal token and selected local secrets, so this warrants a warning. The behavior is runtime extension setup and no concrete malicious action or unconsented install-time mutation was established.
Evidence
dist/launcher.jsdist/mcpPackages.jspackage.json
Network endpoints2
app.ai-agentos.jpregistry.npmjs.org
Decision evidence
public snapshotAI called this Suspicious at 84.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The package exposes CLI launchers, while its manifest has no install-time lifecycle hook; its prepublishOnly script builds the release.
- The launcher fetches authenticated MCP package metadata from the configured portal when run.
- For each returned package, the launcher prepares its executable and places it in the agent's MCP configuration.
- When a package is missing locally, the package manager resolves a version and installs it; the launcher passes the MCP process the portal token and package-declared or selected vault secrets.
Evidence against
- The extension flow is runtime behavior reached by launching the CLI, not an npm install hook.
- The code limits vault-secret forwarding to package-declared or specifically selected names; source inspection did not establish secret exfiltration or another concrete malicious action.
Behavioral surface
CryptoDynamicRequireEnvironmentVarsFilesystemNetwork
Source & flagged code
4 flagged · loading sourcedist/feedback.jsView file
1'use strict';const _0xf0a021=_0xdd3f;(function(_0x18f6af,_0x5902e2){const _0x110e14=_0xdd3f,_0x45e76d=_0x18f6af();while(!![]){try{const _0xed554f=parseInt(_0x110e14(0x13b))/0x1*(-p...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/feedback.jsView on unpkg · L11'use strict';const _0xf0a021=_0xdd3f;(function(_0x18f6af,_0x5902e2){const _0x110e14=_0xdd3f,_0x45e76d=_0x18f6af();while(!![]){try{const _0xed554f=parseInt(_0x110e14(0x13b))/0x1*(-p...
Medium
Dynamic Require
Package source references dynamic require/import behavior.
dist/feedback.jsView on unpkg · L1dist/launcher.jsView file
•matchType = previous_version_dangerous_delta
matchedPackage = @funny-booth/agent-core@0.1.25
matchedIdentity = npm:QGZ1bm55LWJvb3RoL2FnZW50LWNvcmU:0.1.25
similarity = 0.737
summary = stored previous version shares package body but lacks this dangerous source file
Findings
1 Critical3 High4 Medium6 Low
CriticalPrevious Version Dangerous Deltadist/launcher.js
HighSame File Env Network Executiondist/launcher.js
HighObfuscated Payload Loaderdist/feedback.js
HighObfuscated
MediumDynamic Requiredist/feedback.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to @funny-booth/agent-core@0.1.26.
See all versions of @funny-booth/agent-core for other recorded verdicts.
- Review the evidence and your use of @funny-booth/agent-core@0.1.26 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.