registry / @funny-booth/agent-core / 0.1.8

@funny-booth/agent-core@0.1.8

FUNNY BOOTH 専用AIエージェント起動CLI（salesbot1..10 ランチャー + MCPサーバー）

AI Security Review

scanned 58m ago · by lpm-firewall-ai

A user-run launcher accepts portal-controlled MCP package names and starts them through `npx` with local secrets and a portal JWT in their environment. The MCP server also executes a portal-supplied command after sending environment values to that portal.

Static reason

One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source

Trigger

Running a `salesbot*` launcher or invoking the exposed MCP setup/install tool.

Impact

Remote code can execute on the host and access decrypted vault secrets, portal tokens, and supplied environment values.

Mechanism

Portal-controlled package execution and command execution with secret-bearing environment injection.

Rationale

Static source inspection confirms portal-controlled arbitrary execution paths and secret propagation. The absence of an npm install hook does not mitigate the concrete runtime remote-code-execution and credential-exposure behavior. Product guard normalized a non-low false-positive publish_block request to warn-only suspicious.

Evidence

dist/launcher.jsdist/index.jsdist/vault.jslauncher.config.jsonpackage.json

Network endpoints2

prompt-injection-tool-portal.vercel.appregistry.npmjs.org/

@funny-booth/agent-core@0.1.8

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code