No concrete malicious chain was found, but the package intentionally provides high-impact, remotely directed AI-agent execution and mutable remote-code/update paths. This warrants a warning rather than a block.
Static reason
One or more suspicious static signals were detected.
Trigger
User runs acc-runner watch after login; companion persistence requires acc-runner companion install.
Impact
A compromised control plane, task, MCP package, or npm release could execute code with the runner user's repository and agent credentials.
Mechanism
Remote task execution through permissive AI CLIs, temporary MCP configuration, and autonomous package upgrade.
Evidence
package.jsondist/cli.jsdist/task-runner.jsdist/engines/claude-code.jsdist/mcp-spawn.jsdist/runtime/self-upgrade.jsdist/companion-autostart.js<task-workdir>/.mcp.json~/Library/LaunchAgents/com.tokenfactory.acc-runner-companion.plist~/.config/systemd/user/com.tokenfactory.acc-runner-companion.service