@testsmith/api-spector-agent@0.6.1
API Spector private runner: monitor internal/local APIs the cloud cannot reach, with no inbound connections.
AI Security Review
scanned 1mo ago · by lpm-firewall-aiInstall metadata has no lifecycle scripts, and the only runtime network client sends the operator-provided token to the operator-provided APP_URL. Bundled Electron and vm script execution are unused or product-aligned and do not form a concrete attack.
Decision evidence
public snapshot- package.json publishes a CLI named api-spector-agent whose bin and main are dist/index.cjs, with no preinstall, install, or postinstall scripts.
- The agent reads APP_URL, AGENT_TOKEN, and optional RUNTIME_TOKEN from the process environment and refuses to start unless a URL and a token are present.
- Bearer or runtime-token headers are attached only to fetch calls under that caller-supplied APP_URL for the due-monitor feed and result posts.
- A bundled copy of Electron's installer uses spawnSync on install.js, but that path runs only when an in-memory secret store entry exists, and nothing in this package writes that store.
- Due checks can run pre/post scripts in a vm sandbox; that matches the documented private-runner product and is activated only after the user starts the agent against their own cloud.
Source & flagged code
6 flaggedSource contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/index.cjsView on unpkg · L247A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
dist/index.cjsView on unpkgFindings
Affected versions and remediation
This report applies to @testsmith/api-spector-agent@0.6.1.
See all versions of @testsmith/api-spector-agent for other recorded verdicts.
- Review the evidence and your use of @testsmith/api-spector-agent@0.6.1 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.