Only the latest minor receives security updates.
| Version | Supported |
|---|---|
| 0.7.x | ✅ |
| < 0.7 | ❌ |
If you find a security issue in ChatbotLite — for example a way to leak API keys, bypass anti-hallucination guards, exfiltrate conversation state, or smuggle markup through the widget — please do not open a public issue.
Email: security@agents.io
Include:
- A short description of the issue
- A reproduction (smallest possible test case or code snippet)
- Affected version (
chatbotlite@x.y.z) - Your assessment of impact
We will acknowledge within 72 hours and aim to ship a fix or mitigation within 14 days for confirmed high-severity issues.
- The
chatbotlitenpm package and its build artefacts. - The widget UI: input handling, message rendering, tool-card props, streaming parser.
- The
ChatBotserver class: prompt assembly, provider chain, anti-hallucination guards. - Adapters under
chatbotlite/adapters(URL safety, redirect handling).
- Issues that require the customer's own LLM provider keys to be compromised first (e.g. someone with OpenAI access can already do anything).
- Demo deployments on
chatbotlite-demos.vercel.app(demos use a rate-limited shared key for illustration only). - Vulnerabilities in third-party services the customer chooses to wire up (Stripe, Calendly, etc).
- Browser extensions that modify the widget at runtime.
- Never expose your LLM provider keys in client-side code. Put them in your
/api/chathandler only. - Validate the customer's input on your server before forwarding to the LLM.
- Rate-limit your
/api/chatendpoint — the package does not do this for you. - Keep
chatbotliteup to date —npm outdated chatbotlite.