What is serverless-litellm CLAUDE.md?

Deployment instructions for running LiteLLM, an API proxy for language models, on Amazon’s EKS Kubernetes service and connecting it to AWS Bedrock. They cover infrastructure setup, model routing, configuration, and deployment scripts.

Deploying the proxy with Terraform and Kubernetes, configuring model aliases and fallbacks, building its Docker image, and obtaining the service endpoint and API key.

They explain how the project creates and connects its cloud resources, reducing confusion when deploying or troubleshooting the service.

serverless-litellm CLAUDE.md — Instructions file by cncoder

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add instructions/cncoder/serverless-litellm/claude-md

Clone the repo

git clone --depth 1 https://github.com/cncoder/serverless-litellm

Per session 1,124 This file is loaded in full into every session.

When invoked 1,124 The same file — it is already loaded in full.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.01124	$0.01124
Opus 5	$0.00562	$0.00562
Sonnet 5	$0.00225	$0.00225
Haiku 4.5	$0.00112	$0.00112

Measured 2d ago against content hash df31a5ea1a4d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

serverless-litellm CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 118 lines

How it starts

The opening of the file, as written. The whole thing — 118 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — LiteLLM on EKS 项目指南

项目概述

在 AWS EKS 上部署 LiteLLM 代理，让团队通过统一 API Key 使用 Bedrock Claude 模型。

用户 → Claude Code → CloudFront → ALB → LiteLLM (EKS Fargate) → AWS Bedrock

快速部署（一键）

./scripts/setup.sh

交互式引导，约 15-20 分钟完成。自动创建 EKS 集群、RDS、ALB、WAF 等。

前置条件

AWS CLI v2（已配置凭证，需 AdministratorAccess 或等效权限）
Terraform ≥ 1.5
kubectl
Helm 3
envsubst（macOS: brew install gettext）

部署流程

scripts/setup.sh → Terraform 创建基础设施（VPC/EKS/RDS/ECR/WAF）
自动构建 LiteLLM Docker 镜像推送到 ECR
部署 K8s 资源（Deployment/Service/Ingress/HPA/ConfigMap）
输出 CloudFront 域名 + API Key

目录结构

terraform/          Terraform IaC（EKS, VPC, RDS, ECR, WAF, IAM）
kubernetes/         K8s 清单（ConfigMap 含模型路由 + Deployment + Ingress）
  └── configmap.yaml   ← 模型列表、Fallback 链、别名（核心配置）
scripts/            部署脚本 + 测试脚本
docs/               文档（架构/模型/API/排障/监控）
skills/             Claude Code Skills（CloudFront WAF 加固）

核心配置文件

`kubernetes/configmap.yaml`

LiteLLM 的所有行为都在这里：

model_list: 模型路由表（~23 条，含别名和 Fallback）
litellm_settings.drop_params: true: 自动丢弃 Bedrock 不支持的参数
general_settings.master_key: API 主密钥

模型名支持多种格式：claude-sonnet-4-6 / sonnet / claude-sonnet-4-6-20250514 / us.anthropic.claude-sonnet-4-6

`terraform/`

main.tf → EKS 集群 + Fargate Profile
rds.tf → PostgreSQL（API Key 管理 + 用量统计）
waf.tf → WAF 速率限制
variables.tf → 所有可配置项
terraform.tfvars 绝不提交（已在 .gitignore）

常见开发任务

添加新模型

编辑 kubernetes/configmap.yaml，在 model_list 下新增：

- model_name: "新模型短名"
  litellm_params:
    model: "bedrock/us.anthropic.模型ID"
    aws_region_name: "us-east-1"

然后 kubectl apply -f kubernetes/configmap.yaml && kubectl rollout restart deployment litellm-deployment -n litellm

更新文档

用户指南: docs/claude-code.md（CC 配置、排障、迁移）
模型列表: docs/models.md
README: README.md（中文）/ README.en.md（英文），保持同步

测试

# API 级别测试（需要 CloudFront 域名 + API Key）
bash scripts/test-models.sh

# E2E 测试
bash scripts/e2e-test.sh

Read the full file on GitHub · 118 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2d ago First seen · 118 lines · 1,124 tokens per session scan A df31a5ea1a4d

Subscribe to this mod's changes

serverless-litellm CLAUDE.md is an instructions file published in the GitHub repository cncoder/serverless-litellm (2 stars, last pushed 1mo ago), licensed MIT. It adds 1,124 tokens to every session, about $0.0056 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Backend and APIs Data and AI DevOps and cloud ai-gateway aws bedrock claude claude-code eks

serverless-litellm CLAUDE.md