autopilot-ai@1.20.1
autopilot (formerly kimiflare): a terminal coding agent that runs any OpenRouter model with your own key. Per-turn billed cost, web search, GitHub, headless browser, LSP, MCP, local memory.
AI Security Review
scanned 24d ago · by lpm-firewall-aiThe manifest has no install-time mutation hook, and the flagged agent-context locations read bounded project files rather than altering agent configuration. The inspected credential routing and network behavior are package-aligned; scanner hints alone do not establish a concrete attack.
Decision evidence
public snapshot- package.json exposes a CLI and two library entrypoints; its only lifecycle hook is prepublishOnly building the package.
- The flagged agent-context code reads a bounded set of project context files; it does not show a write to agent configuration or a plugin install.
- The SDK bundle uses the same bounded context-file reader.
- The declaration file contains type declarations rather than executable behavior.
- Requesty endpoint selection requires HTTPS and restricts hostnames to Requesty domains.
- OpenRouter API keys are sent in the Authorization header to the OpenRouter service; Requesty keys are used only when no custom endpoint or OpenRouter key is selected.
- A feedback-worker URL uses the package's kimiflare.com domain.
Source & flagged code
14 flaggedSource appears to send environment or credential material through DNS lookups.
dist/index.jsView on unpkg · L59Source passes code obtained from a remote response into a dynamic execution sink.
dist/index.jsView on unpkg · L59Findings
Affected versions and remediation
This report applies to autopilot-ai@1.20.1.
See all versions of autopilot-ai for other recorded verdicts.
- Review the evidence and your use of autopilot-ai@1.20.1 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.