pennyrouter@0.3.4
Install and manage PennyRouter local coding-agent integrations.
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThis is an explicit-user-command agent integration, not an install-time attack. Its unpinned npx persistence and agent-control-surface changes are a concrete ongoing risk, so it should be warned rather than blocked.
Decision evidence
public snapshot- The install and update flows register an MCP server in existing Claude Code and Codex configuration.
- That MCP entry runs an unpinned `npx -y pennyrouter mcp`, allowing a later registry release to execute when the agent starts it.
- The Claude status-line integration replaces the configured command and executes a previously configured command through a shell.
- The CLI sends user-supplied upstream credentials to the selected gateway.
- package.json has no preinstall, install, or postinstall lifecycle hook.
- The configuration changes occur only after the user runs the CLI install or update command.
- MCP writers back up existing files and avoid overwriting an unrelated server with the same name.
- The CodeGPT harness only provides manual instructions and clipboard support.
Source & flagged code
6 flagged · loading sourceSource writes installer persistence such as shell profile or service configuration.
src/launch.test.jsView on unpkg · L1Findings
Affected versions and remediation
This report applies to pennyrouter@0.3.4.
See all versions of pennyrouter for other recorded verdicts.
- Review the evidence and your use of pennyrouter@0.3.4 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.