mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-02 02:11:58 +00:00
Adds a standalone MCP server (provisioning_mcp/) that lets authenticated AI agents provision ephemeral LiteLLM deployments from the helm/litellm chart for end-to-end testing. Given a repo URL and git revision it derives the component image repos, optionally stands up throwaway in-cluster Postgres/Redis, and runs helm upgrade --install in the litellm namespace. - OAuth 2.0 resource-server auth: validates JWT bearer tokens against the issuer's JWKS (signature, iss, aud, exp, required scope). - Tools: provision / delete / status / list deployments. - Dockerfile pins + checksum-verifies helm and kubectl; namespaced RBAC and Deployment manifests under deploy/. - Unit tests cover auth, naming/registry derivation, manifests, and the provisioning flow (35 tests). https://claude.ai/code/session_019WDsdJNGjNyUNrso3xigSV |
||
|---|---|---|
| .. | ||
| deploy | ||
| src/litellm_provisioning_mcp | ||
| tests | ||
| .gitignore | ||
| Dockerfile | ||
| pyproject.toml | ||