@companion-ai/feynman@0.5.56
The open source AI research agent.
AI Security Review
scanned 12d ago · by lpm-firewall-aiSource inspection found a Node-version install guard, user-triggered project setup and AlphaXiv authentication, and package-aligned runtime telemetry. Although failed commands may send up to 4 KB of Pi stderr, the telemetry has an opt-out and does not establish malicious behavior.
Decision evidence
public snapshot- The package’s only install hook runs a Node version check; the script compares the installed version with the minimum and exits when it is too old.
- The flagged project-file write belongs to the explicit `/init` command. It creates `AGENTS.md` only when that file is absent and preserves an existing one.
- AlphaXiv login is triggered by an explicit command or setup confirmation and uses the AlphaXiv authentication issuer.
- The inspected dependencies provide prompt UI, AlphaXiv authentication, and XML parsing functionality through their declared entrypoints.
- Runtime telemetry reports command metadata and may include up to 4 KB of Pi stderr after a failed command. The package displays an opt-out notice; no unrelated credential recipient or agent-control mutation was found.
- Telemetry is enabled by default and may send Pi stderr from failed commands to the configured or default PostHog host.
Source & flagged code
5 flaggedPackage defines install-time lifecycle scripts.
package.jsonView on unpkgSource gates dangerous network, credential, or execution behavior behind CI, host, platform, time, or geo fingerprint checks.
bin/feynman.jsView on unpkg · L23Findings
Affected versions and remediation
This report applies to @companion-ai/feynman@0.5.56.
See all versions of @companion-ai/feynman for other recorded verdicts.
- Review the evidence and your use of @companion-ai/feynman@0.5.56 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.