npm

biklirouter @0.5.47

Vulnerability report · Last retrieved from osv.dev August 13, 2026 at 4:27 AM UTC

Malicious

OSV ID

MAL-2026-13787

Ecosystem

npm

Summary

Package ships a compiled Next.js application (server routes under app/.next-cli-build/server/app/api/) implementing an LLM/AI-provider router: OAuth import flows for provider CLIs (codex, cursor, cline, kilo, droid, hermes, jcode, deepseek, opencode, openclaw, grok, devin, cowork), OpenAI-compatible endpoints (/api/v1/chat/completions, /embeddings, /images/generations, /videos/*, /responses, /messages), provider validation/testing (ping, curl), and Tailscale tunnel management. Many API route bundles include child_process, os, http/https, and process.env because Next.js server bundling co-locates these Node built-ins per route file; static keyword co-occurrence of require('child_process') + POST + process.env in minified route bundles matches the exfiltration pattern shape but reflects normal Next.js server-side rendering plus the app's documented purpose (spawn CLI tools, validate providers via ping, test provider endpoints). cli.js performs a version check against https://registry.npmjs.org (the official npm registry) and reads os.homedir() for local config. No traced code path shows installer-secret paths (~/.aws, ~/.ssh, ~/.npmrc contents) being read and sent to an author-controlled host; no lifecycle scripts fetch-and-exec remote content; no hardcoded non-registry C2 destination is confirmed in the evidence. The behavior surface (spawning CLIs, reading provider credentials that the user configures into the app, contacting external LLM providers) is consistent with the app's declared router purpose, but the density of shell-exec + network primitives across dozens of route handlers plus a MITM server module (app/src/mitm/server.js) warrants human review of the maintainer and endpoints before trust.

Source: amazon-inspector (aa35b0b8fe046a8dd43a4307742214223c3f908b5ba65fa6d24dae18b33cff9b)

Protect your entire dependency tree

Scan your lock files automatically on every PR. Block malicious packages before they reach production.