
from litellm-skills35
Guides adding and testing a new model on a LiteLLM proxy: pick provider, supply credentials, POST to /model/new and verify routing with a test call.
Walks an operator through registering a new model with a live LiteLLM proxy. The skill describes required inputs (proxy base URL and admin API key), maps provider options to litellm_params values (OpenAI, Azure, Anthropic, Bedrock, Vertex, Ollama, etc.), and provides example curl commands to create the model and verify it with a test chat request.
Use when you need to add or register a new backend model to a LiteLLM proxy instance, especially during deployment or operator tooling workflows. It's intended for admins who manage model routing and want a safe test step to confirm requests route correctly.
/model/new and test with /chat/completions.model_id and test result.Operator-facing automation agents and CLI-capable helpers that can perform authenticated HTTP requests (agents with curl/http tools).
This skill guides adding a new model to a live LiteLLM proxy instance. It walks through picking a provider, entering credentials, POSTing to /model/new, and verifying with a test call. No bundled scripts — all instructions are curl commands the agent runs directly. The SKILL.md is well-structured with a clear provider table and step-by-step flow. Security is reasonable: credentials are prompted from the user rather than hardcoded, but the skill doesn't warn about credential exposure in shell history or suggest masking API keys in output.
Clean, straightforward skill from BerriAI's official repo. Well-written provider table is genuinely useful. Minor security concern: no guidance on credential hygiene (shell history, log masking). Architecture is solid but lean — no scripts/ or references/ directories, just a single SKILL.md with inline commands.
LiteLLM Agent Creator
Programmatically create and configure new AI agents on a live LiteLLM proxy using the /v1/agents endpoint.
Add Team — LiteLLM Proxy
Create and configure a new team on a LiteLLM proxy: collects team name, budget, allowed models and calls the proxy API to provision the team, returning the team
LiteLLM — Update Team
Update an existing team on a LiteLLM proxy: change alias, budget, allowed models, and rate limits via the proxy API.
Add User (LiteLLM Proxy)
Interactive helper to create a new user on a LiteLLM proxy by collecting email, role, budget and allowed models, then calling the proxy's user creation API.
Add Organization (LiteLLM)
Create a new organization on a LiteLLM proxy: prompts for org name, allowed models, budget and calls the proxy API to create the org.
Delete Organization
Delete one or more organizations from a LiteLLM proxy with confirmation and safe checks.
Delete MCP Server
Interactively delete an MCP server registration from a LiteLLM proxy: ask for server_id, confirm, and call the DELETE endpoint safely.
LiteLLM: Add API Key
Create and provision API keys on a live LiteLLM proxy, specifying alias, scope, allowed models, budget and expiry.
Update User (LiteLLM)
Update an existing user on a LiteLLM proxy: change budget, role, models, and rate limits via the proxy's POST /user/update endpoint.
Delete API Key (LiteLLM)
Remove one or more API keys or key aliases from a LiteLLM proxy after confirming with the user; calls POST /key/delete and returns the deleted_keys list.