TeamPCP Backdoors LiteLLM Versions 1.82.7-1.82.8 via Trivy CI/CD Compromise

Threat actor TeamPCP compromised LiteLLM versions 1.82.7-1.82.8 by poisoning Trivy in a CI/CD supply chain attack. Here is what developers need to know.
TeamPCP Plants a Backdoor Inside LiteLLM Through CI/CD Poisoning
A threat actor operating under the handle TeamPCP successfully backdoored LiteLLM versions 1.82.7 and 1.82.8 by compromising the project's CI/CD pipeline through a tampered Trivy integration. If you are running either of those versions in production, treat your environment as compromised until proven otherwise.
What Happened
LiteLLM is a popular open-source proxy library used by developers to interface with dozens of LLM APIs - OpenAI, Anthropic, Cohere, you name it. Its widespread adoption in AI-heavy stacks made it a high-value target.
Here is the attack chain as understood:
- Trivy, the open-source vulnerability scanner commonly embedded in CI/CD pipelines, was the entry point
- TeamPCP manipulated a Trivy dependency or configuration within LiteLLM's build workflow
- This allowed malicious code to be injected silently into the published package artifacts for versions 1.82.7 and 1.82.8
- The backdoor rode the automated release pipeline straight into the official distribution without triggering standard review gates
This is a textbook CI/CD supply chain attack - not a zero-day in LiteLLM's application code, but a compromise of the trust layer that builds and ships it.
Why This Is a Big Deal
Any developer or organization that pulled litellm==1.82.7 or litellm==1.82.8 via pip is potentially running attacker-controlled code inside their LLM gateway. That means:
- API keys and model credentials are at risk
- Proxied requests to LLM providers could be intercepted or exfiltrated
- Internal network access could be exposed depending on deployment context
How to Protect Yourself Right Now
- Immediately pin or downgrade to a known-safe version (1.82.6 or earlier until a clean release is confirmed)
- Audit your pipeline tools - Trivy and similar scanners have privileged access; treat them like production code
- Verify package checksums against official releases before deploying
- Rotate all API keys and secrets that were accessible to the LiteLLM process
- Review CI/CD permissions - scanners and build tools should operate with least-privilege, never write access to release artifacts
- Enable dependency pinning with hash verification in your
requirements.txtorpyproject.toml
Harden Your Supply Chain
The LiteLLM incident is a reminder that the weakest link in modern software delivery is often not your application code - it is the automated scaffolding around it. Your scanner, your linter, your test runner - all of them are potential pivot points if left unsecured.
Is your app vulnerable to similar attacks? Run an automated scan in 3 minutes with VibeShield.
Free security scan
Is your app vulnerable to similar attacks?
VibeWShield automatically scans for these and 18 other security checks in under 3 minutes.
Scan your app free