traicebox 0.1.0 → 0.1.1

package/README.md

@@ -0,0 +1,256 @@
+# Traicebox
+
+**Traicebox** is a zero-config local developer stack for tracing and session tracking around LLM and AI model workflows. It enables developers to have a working local tracing and inspection setup within minutes.
+
+### Key Use Cases
+
+- **Tool & Prompt Development**: Debug plugins, skills, or prompts for harness tools like [OpenCode](https://opencode.ai/) with full visibility.
+- **Model Evaluation**: Evaluate and compare local models with detailed trace logging. The built-in config generator makes setting up tools remarkably simple.
+- **Application Development**: Build AI applications with minimal code changes. Requests routed through the built-in LiteLLM proxy automatically generate traces in Langfuse. You can group these traces into sessions by adding an `x-litellm-session-id` header to your requests.
+
+### Included in the Stack
+
+- **LiteLLM**: Unified proxy for LLM backends with custom session-tracking callbacks.
+- **Langfuse**: Open-source tracing and observability for LLM applications.
+- **Auto-login Proxies**: Dynamic proxies providing immediate, authenticated access to LiteLLM UI and Langfuse.
+- **Harness Integration**: Automatic configuration generation for external tools like OpenCode.
+
+## Prerequisites
+
+Traicebox requires [Docker](https://www.docker.com/) and [Docker Compose](https://docs.docker.com/compose/) to be installed and running on your system.
+
+## Installation
+
+### Published Package (Recommended)
+
+You can install Traicebox via NPM or Bun:
+
+```bash
+# Using Bun
+bun install -g traicebox
+
+# Using NPM
+npm install -g traicebox
+```
+
+### Binary Release
+
+Download the latest prebuilt binary for your platform from the [GitHub Releases](https://github.com/fardjad/traicebox/releases) page.
+
+Once downloaded, make it executable and move it to your path:
+
+```bash
+chmod +x traicebox-macos-arm64
+mv traicebox-macos-arm64 /usr/local/bin/traicebox
+```
+
+## Shell Completion
+
+Traicebox supports shell completion for Bash and Zsh.
+
+### Zsh
+
+Add the following to your `~/.zshrc`:
+
+```bash
+# Enable completion if not already enabled
+autoload -Uz compinit && compinit
+
+source <(traicebox completion)
+```
+
+### Bash
+
+Add the following to your `~/.bashrc`:
+
+```bash
+source <(traicebox completion)
+```
+
+## Usage
+
+### 1. Initial Setup
+
+Initialize the Traicebox home directory:
+
+```bash
+traicebox setup
+```
+
+### 2. Import Models
+
+Before starting the stack, you need to import your available models into LiteLLM. Traicebox requires an endpoint URL or alias to fetch models from.
+
+Supported aliases for common local tools:
+
+| Alias       | Default Endpoint URL               |
+| :---------- | :--------------------------------- |
+| `lm-studio` | `http://127.0.0.1:1234/v1/models`  |
+| `ollama`    | `http://localhost:11434/v1/models` |
+| `llama-cpp` | `http://localhost:8080/v1/models`  |
+
+```bash
+traicebox models import-from-openai-api --endpoint lm-studio
+```
+
+To use a custom OpenAI-compatible endpoint URL:
+
+```bash
+traicebox models import-from-openai-api --endpoint http://your-api:port/v1/models
+```
+
+If the endpoint requires authentication, provide the API key via the `OPENAI_COMPATIBLE_API_KEY` environment variable:
+
+```bash
+OPENAI_COMPATIBLE_API_KEY="your-api-key" traicebox models import-from-openai-api --endpoint http://your-api:port/v1/models
+```
+
+### 3. Configure Harness Integration (Optional)
+
+If you use [OpenCode](https://opencode.ai/), you can synchronize its model configuration with Traicebox.
+
+> [!WARNING]
+> The following command will overwrite your existing OpenCode configuration file.
+
+```bash
+traicebox generate-harness-config opencode > ~/.config/opencode/opencode.jsonc
+```
+
+### 4. Start the Stack
+
+```bash
+traicebox start
+```
+
+Once running, you can access the public endpoints:
+
+- **LiteLLM**: `http://litellm.localhost:5483`
+- **Langfuse**: `http://langfuse.localhost:5483`
+
+### 5. Verify and Test
+
+If you configured OpenCode in step 3, you can test the integration:
+
+```bash
+# Refresh the models cache
+opencode models --refresh
+
+# Run a test prompt
+opencode run -m 'litellm/your/model' 'Just write a greeting message!'
+```
+
+> [!NOTE]
+> Replace `'litellm/your/model'` with one of the models imported in step 2.
+
+After running a prompt, you can inspect the captured traces in Langfuse by visiting:
+[http://langfuse.localhost:5483/project/local-project/sessions](http://langfuse.localhost:5483/project/local-project/sessions)
+
+There, you can select the latest session to inspect system prompts, user messages, and tool interactions.
+
+#### Application Integration (with Session Tracking)
+
+You can integrate Traicebox into your own applications by pointing your LLM client to the LiteLLM proxy. To group related traces together, simply include an `x-litellm-session-id` header in your requests.
+
+Here is an example using `curl` to demonstrate session grouping:
+
+```bash
+MODEL_NAME="your-model-name"
+
+# First request in a session
+curl http://litellm.localhost:5483/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer sk-litellm-local-client" \
+  -H "x-litellm-session-id: my-test-session" \
+  -d "{
+    \"model\": \"$MODEL_NAME\",
+    \"messages\": [{\"role\": \"user\", \"content\": \"Hello! This is a test message to start my session.\"}]
+  }"
+
+# Second request in the same session
+curl http://litellm.localhost:5483/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer sk-litellm-local-client" \
+  -H "x-litellm-session-id: my-test-session" \
+  -d "{
+    \"model\": \"$MODEL_NAME\",
+    \"messages\": [{\"role\": \"user\", \"content\": \"Goodbye! I am done with this test session.\"}]
+  }"
+```
+
+> [!NOTE]
+> Set `MODEL_NAME` to one of the exact model names imported in step 2. The default API key is `sk-litellm-local-client`.
+
+Refresh the [Langfuse sessions page](http://langfuse.localhost:5483/project/local-project/sessions) to see both traces grouped under the `my-test-session` ID.
+
+### Other Useful Commands
+
+| Command                  | Description                                     |
+| :----------------------- | :---------------------------------------------- |
+| `traicebox stop`         | Stop the stack.                                 |
+| `traicebox restart`      | Recreate and restart the stack.                 |
+| `traicebox destroy`      | Remove the stack and delete local data volumes. |
+| `traicebox models clear` | Clear the custom model list from LiteLLM.       |
+
+For more information on available commands and options, run `traicebox --help`.
+
+## Configuration
+
+Traicebox stores its configuration and data in `${TRAICEBOX_HOME}`. You can override this location by setting the `TRAICEBOX_HOME` environment variable.
+
+By default, it is located at:
+
+| OS                 | Default `${TRAICEBOX_HOME}`                            |
+| :----------------- | :----------------------------------------------------- |
+| **macOS**          | `~/Library/Application Support/Traicebox`              |
+| **Windows**        | `%APPDATA%\Traicebox`                                  |
+| **Linux / Others** | `~/.config/traicebox` (or respects `$XDG_CONFIG_HOME`) |
+
+### `traicebox.yaml`
+
+You can customize the stack behavior by creating or editing `${TRAICEBOX_HOME}/traicebox.yaml`.
+
+| Option | Type     | Default     | Description                                                  |
+| :----- | :------- | :---------- | :----------------------------------------------------------- |
+| `host` | `string` | `127.0.0.1` | The host address that Traicebox services will bind to.       |
+| `port` | `number` | `5483`      | The port that Traicebox services will be accessible through. |
+
+Example:
+
+```yaml
+host: 127.0.0.1
+port: 5483
+```
+
+## LiteLLM
+
+LiteLLM loads its proxy config from `${TRAICEBOX_HOME}/litellm/config.yaml`. The LiteLLM UI is available at `http://litellm.localhost:5483`, where Traicebox automatically manages your admin session for immediate access.
+
+`${TRAICEBOX_HOME}/litellm/config.yaml` is the source of truth for model routing and upstream API wiring. Imported models are written directly into `model_list`, with `api_base` inferred from the import endpoint and `api_key` read from `os.environ/OPENAI_COMPATIBLE_API_KEY`.
+
+`OPENAI_COMPATIBLE_API_KEY` is intentionally not persisted by Tr**ai**cebox. For local backends such as LM Studio, Ollama, or llama.cpp, no secret is needed. For authenticated upstreams, run `traicebox` through your password manager CLI so it injects `OPENAI_COMPATIBLE_API_KEY` into the process; Tr**ai**cebox will materialize that into an ephemeral Docker secret for LiteLLM only.
+
+To replace the LiteLLM `model_list` from an OpenAI-compatible `/v1/models` endpoint (or alias like `lm-studio`, `ollama`, `llama-cpp`):
+
+```bash
+traicebox models import-from-openai-api --endpoint lm-studio
+```
+
+If that endpoint requires auth, inject `OPENAI_COMPATIBLE_API_KEY` into the command environment first. If it does not, the command works without a key.
+
+To clear the LiteLLM `model_list`:
+
+```bash
+traicebox models clear
+```
+
+To print an OpenCode config snippet for the current LiteLLM models:
+
+```bash
+traicebox generate-harness-config opencode
+```
+
+## Langfuse
+
+LiteLLM sends OTEL traces to Langfuse using the built-in `langfuse_otel` callback and `LANGFUSE_OTEL_HOST=http://langfuse-web:3000`.
+
+Langfuse is accessible at `http://langfuse.localhost:5483`. By default, Traicebox disables public signups and automatically creates a browser session for the seeded admin user so you can start tracing immediately.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "traicebox",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "description": "A zero-config local developer stack for tracing and session tracking around LLM and AI model workflows",
   "license": "MIT",
   "author": "Fardjad Davari <public@fardjad.com>",
@@ -47,4 +47,4 @@
     "yaml": "^2.8.3",
     "yargs": "^18.0.0"
   }
-}
+}

package/dist/00-app-databases-mx6j59m0.sh

@@ -1,28 +0,0 @@
-#!/bin/sh
-set -eu
-
-psql -v ON_ERROR_STOP=1 --username "$POSTGRES_USER" --dbname "$POSTGRES_DB" <<SQL
-DO \$\$
-BEGIN
-  IF NOT EXISTS (SELECT FROM pg_roles WHERE rolname = '${LITELLM_DB_USER}') THEN
-    CREATE ROLE ${LITELLM_DB_USER} LOGIN PASSWORD '${LITELLM_DB_PASSWORD}';
-  END IF;
-END
-\$\$;
-
-DO \$\$
-BEGIN
-  IF NOT EXISTS (SELECT FROM pg_roles WHERE rolname = '${LANGFUSE_DB_USER}') THEN
-    CREATE ROLE ${LANGFUSE_DB_USER} LOGIN PASSWORD '${LANGFUSE_DB_PASSWORD}';
-  END IF;
-END
-\$\$;
-
-SELECT 'CREATE DATABASE ${LITELLM_DB_NAME} OWNER ${LITELLM_DB_USER}'
-WHERE NOT EXISTS (SELECT FROM pg_database WHERE datname = '${LITELLM_DB_NAME}')
-\gexec
-
-SELECT 'CREATE DATABASE ${LANGFUSE_DB_NAME} OWNER ${LANGFUSE_DB_USER}'
-WHERE NOT EXISTS (SELECT FROM pg_database WHERE datname = '${LANGFUSE_DB_NAME}')
-\gexec
-SQL

package/dist/Caddyfile-at2nzhxs.

@@ -1,18 +0,0 @@
-{
-  auto_https off
-}
-
-http://langfuse.localhost:{$TRAICEBOX_PORT} {
-  reverse_proxy langfuse-proxy:3000
-}
-
-http://litellm.localhost:{$TRAICEBOX_PORT} {
-  @litellm_ui path /ui /ui/*
-  handle @litellm_ui {
-    reverse_proxy litellm-ui-proxy:4000
-  }
-
-  handle {
-    reverse_proxy litellm:4000
-  }
-}

package/dist/Dockerfile-bzexf8bh.

@@ -1,7 +0,0 @@
-FROM oven/bun:1
-
-WORKDIR /app
-
-COPY langfuse-proxy/server.ts /app/server.ts
-
-CMD ["bun", "run", "/app/server.ts"]

package/dist/Dockerfile-tvb2c6ma.

@@ -1,7 +0,0 @@
-FROM oven/bun:1
-
-WORKDIR /app
-
-COPY litellm-ui-proxy/server.ts /app/server.ts
-
-CMD ["bun", "run", "/app/server.ts"]

package/dist/bootstrap-client-key-v79zzaxg.sh

@@ -1,39 +0,0 @@
-#!/bin/sh
-set -eu
-
-LITELLM_URL="http://litellm:4000"
-READY_URL="$LITELLM_URL/health/readiness"
-KEY_INFO_URL="$LITELLM_URL/key/info?key=${LITELLM_CLIENT_KEY}"
-KEY_GENERATE_URL="$LITELLM_URL/key/generate"
-
-for _ in $(seq 1 60); do
-  if curl -fsS "$READY_URL" >/dev/null 2>&1; then
-    break
-  fi
-  sleep 2
-done
-
-if ! curl -fsS "$READY_URL" >/dev/null 2>&1; then
-  echo "LiteLLM did not become ready in time" >&2
-  exit 1
-fi
-
-if curl -fsS \
-  -H "Authorization: Bearer ${LITELLM_MASTER_KEY}" \
-  "$KEY_INFO_URL" >/dev/null 2>&1; then
-  echo "LiteLLM client key already exists"
-  exit 0
-fi
-
-curl -fsS \
-  -X POST \
-  -H "Authorization: Bearer ${LITELLM_MASTER_KEY}" \
-  -H "Content-Type: application/json" \
-  "$KEY_GENERATE_URL" \
-  -d "{
-    \"key_alias\": \"local-client\",
-    \"key\": \"${LITELLM_CLIENT_KEY}\",
-    \"models\": []
-  }" >/dev/null
-
-echo "LiteLLM client key created"

package/dist/compose-1b8sxndd.yml

@@ -1,266 +0,0 @@
-services:
-  db:
-    image: pgautoupgrade/pgautoupgrade:18-alpine
-    environment:
-      - POSTGRES_DB=postgres
-      - POSTGRES_USER=${POSTGRES_SUPERUSER}
-      - POSTGRES_PASSWORD=${POSTGRES_SUPERPASS}
-      - PGAUTO_DEVEL=
-      - PGAUTO_ONESHOT=no
-      - LITELLM_DB_NAME=${LITELLM_DB_NAME}
-      - LITELLM_DB_USER=${LITELLM_DB_USER}
-      - LITELLM_DB_PASSWORD=${LITELLM_DB_PASSWORD}
-      - LANGFUSE_DB_NAME=${LANGFUSE_DB_NAME}
-      - LANGFUSE_DB_USER=${LANGFUSE_DB_USER}
-      - LANGFUSE_DB_PASSWORD=${LANGFUSE_DB_PASSWORD}
-    healthcheck:
-      test: ["CMD-SHELL", "pg_isready -U ${POSTGRES_SUPERUSER} -d postgres && psql -U ${POSTGRES_SUPERUSER} -d postgres -tAc \"SELECT 1 FROM pg_database WHERE datname IN ('${LITELLM_DB_NAME}','${LANGFUSE_DB_NAME}')\" | grep -q 1"]
-      interval: 3s
-      timeout: 3s
-      retries: 20
-      start_period: 5s
-    volumes:
-      - postgres_data:/var/lib/postgresql
-      - ./postgres/init/00-app-databases.sh:/docker-entrypoint-initdb.d/00-app-databases.sh:ro
-  langfuse-clickhouse:
-    image: clickhouse/clickhouse-server
-    user: "101:101"
-    environment:
-      - CLICKHOUSE_DB=${CLICKHOUSE_DB}
-      - CLICKHOUSE_USER=${CLICKHOUSE_USER}
-      - CLICKHOUSE_PASSWORD=${CLICKHOUSE_PASSWORD}
-      - TZ=UTC
-    healthcheck:
-      test: ["CMD", "wget", "--no-verbose", "--tries=1", "--spider", "http://localhost:8123/ping"]
-      interval: 5s
-      timeout: 5s
-      retries: 10
-      start_period: 1s
-    volumes:
-      - langfuse_clickhouse_data:/var/lib/clickhouse
-      - langfuse_clickhouse_logs:/var/log/clickhouse-server
-  langfuse-redis:
-    image: redis:latest
-    command:
-      - --requirepass
-      - ${LANGFUSE_REDIS_PASSWORD}
-      - --maxmemory-policy
-      - noeviction
-    healthcheck:
-      test: ["CMD", "redis-cli", "-a", "${LANGFUSE_REDIS_PASSWORD}", "ping"]
-      interval: 3s
-      timeout: 10s
-      retries: 10
-    volumes:
-      - langfuse_redis_data:/data
-  langfuse-minio:
-    image: cgr.dev/chainguard/minio
-    entrypoint: sh
-    command:
-      - -c
-      - mkdir -p /data/langfuse && minio server --address ":9000" --console-address ":9001" /data
-    environment:
-      - MINIO_ROOT_USER=${MINIO_ROOT_USER}
-      - MINIO_ROOT_PASSWORD=${MINIO_ROOT_PASSWORD}
-    healthcheck:
-      test: ["CMD", "mc", "ready", "local"]
-      interval: 1s
-      timeout: 5s
-      retries: 5
-      start_period: 1s
-    volumes:
-      - langfuse_minio_data:/data
-  langfuse-worker:
-    image: langfuse/langfuse-worker:3
-    depends_on:
-      db:
-        condition: service_healthy
-      langfuse-web:
-        condition: service_healthy
-      langfuse-minio:
-        condition: service_healthy
-      langfuse-redis:
-        condition: service_healthy
-      langfuse-clickhouse:
-        condition: service_healthy
-    environment: &langfuse-env
-      NEXTAUTH_URL: http://langfuse.localhost:${TRAICEBOX_PORT}
-      DATABASE_URL: postgresql://${LANGFUSE_DB_USER}:${LANGFUSE_DB_PASSWORD}@db:5432/${LANGFUSE_DB_NAME}
-      SALT: ${LANGFUSE_SALT}
-      ENCRYPTION_KEY: "${LANGFUSE_ENCRYPTION_KEY}"
-      TELEMETRY_ENABLED: "${LANGFUSE_TELEMETRY_ENABLED}"
-      LANGFUSE_ENABLE_EXPERIMENTAL_FEATURES: "${LANGFUSE_EXPERIMENTAL_FEATURES}"
-      CLICKHOUSE_MIGRATION_URL: clickhouse://langfuse-clickhouse:9000
-      CLICKHOUSE_URL: http://langfuse-clickhouse:8123
-      CLICKHOUSE_USER: ${CLICKHOUSE_USER}
-      CLICKHOUSE_PASSWORD: ${CLICKHOUSE_PASSWORD}
-      CLICKHOUSE_CLUSTER_ENABLED: "false"
-      LANGFUSE_S3_EVENT_UPLOAD_BUCKET: langfuse
-      LANGFUSE_S3_EVENT_UPLOAD_REGION: auto
-      LANGFUSE_S3_EVENT_UPLOAD_ACCESS_KEY_ID: ${MINIO_ROOT_USER}
-      LANGFUSE_S3_EVENT_UPLOAD_SECRET_ACCESS_KEY: ${MINIO_ROOT_PASSWORD}
-      LANGFUSE_S3_EVENT_UPLOAD_ENDPOINT: http://langfuse-minio:9000
-      LANGFUSE_S3_EVENT_UPLOAD_FORCE_PATH_STYLE: "true"
-      LANGFUSE_S3_EVENT_UPLOAD_PREFIX: events/
-      LANGFUSE_S3_MEDIA_UPLOAD_BUCKET: langfuse
-      LANGFUSE_S3_MEDIA_UPLOAD_REGION: auto
-      LANGFUSE_S3_MEDIA_UPLOAD_ACCESS_KEY_ID: ${MINIO_ROOT_USER}
-      LANGFUSE_S3_MEDIA_UPLOAD_SECRET_ACCESS_KEY: ${MINIO_ROOT_PASSWORD}
-      LANGFUSE_S3_MEDIA_UPLOAD_ENDPOINT: http://langfuse-minio:9000
-      LANGFUSE_S3_MEDIA_UPLOAD_FORCE_PATH_STYLE: "true"
-      LANGFUSE_S3_MEDIA_UPLOAD_PREFIX: media/
-      REDIS_HOST: langfuse-redis
-      REDIS_PORT: "6379"
-      REDIS_AUTH: ${LANGFUSE_REDIS_PASSWORD}
-  langfuse-web:
-    image: langfuse/langfuse:3
-    depends_on:
-      db:
-        condition: service_healthy
-      langfuse-minio:
-        condition: service_healthy
-      langfuse-redis:
-        condition: service_healthy
-      langfuse-clickhouse:
-        condition: service_healthy
-    environment:
-      <<: *langfuse-env
-      NEXTAUTH_SECRET: ${LANGFUSE_NEXTAUTH_SECRET}
-      AUTH_DISABLE_SIGNUP: ${LANGFUSE_AUTH_DISABLE_SIGNUP}
-      LANGFUSE_INIT_ORG_ID: ${LANGFUSE_INIT_ORG_ID}
-      LANGFUSE_INIT_ORG_NAME: ${LANGFUSE_INIT_ORG_NAME}
-      LANGFUSE_INIT_PROJECT_ID: ${LANGFUSE_INIT_PROJECT_ID}
-      LANGFUSE_INIT_PROJECT_NAME: ${LANGFUSE_INIT_PROJECT_NAME}
-      LANGFUSE_INIT_PROJECT_PUBLIC_KEY: ${LANGFUSE_INIT_PROJECT_PUBLIC_KEY}
-      LANGFUSE_INIT_PROJECT_SECRET_KEY: ${LANGFUSE_INIT_PROJECT_SECRET_KEY}
-      LANGFUSE_INIT_USER_EMAIL: ${LANGFUSE_INIT_USER_EMAIL}
-      LANGFUSE_INIT_USER_NAME: ${LANGFUSE_INIT_USER_NAME}
-      LANGFUSE_INIT_USER_PASSWORD: ${LANGFUSE_INIT_USER_PASSWORD}
-    healthcheck:
-      test: ["CMD", "node", "-e", "const host = process.env.HOSTNAME || 'localhost'; fetch('http://' + host + ':3000/api/auth/csrf').then(async (r) => { const ok = r.ok && (await r.text()).includes('csrfToken'); process.exit(ok ? 0 : 1); }).catch(() => process.exit(1))"]
-      interval: 5s
-      timeout: 5s
-      retries: 24
-      start_period: 20s
-  langfuse-proxy:
-    build:
-      context: .
-      dockerfile: langfuse-proxy/Dockerfile
-    depends_on:
-      langfuse-web:
-        condition: service_healthy
-    environment:
-      - PORT=3000
-      - LANGFUSE_PUBLIC_ORIGIN=http://langfuse.localhost:${TRAICEBOX_PORT}
-      - LANGFUSE_UPSTREAM_ORIGIN=http://langfuse-web:3000
-      - LANGFUSE_PROXY_AUTOLOGIN=${LANGFUSE_PROXY_AUTOLOGIN}
-      - LANGFUSE_INIT_USER_EMAIL=${LANGFUSE_INIT_USER_EMAIL}
-      - LANGFUSE_INIT_USER_PASSWORD=${LANGFUSE_INIT_USER_PASSWORD}
-    healthcheck:
-      test: ["CMD", "bun", "-e", "fetch('http://127.0.0.1:3000/readyz').then((r) => process.exit(r.ok ? 0 : 1)).catch(() => process.exit(1))"]
-      interval: 5s
-      timeout: 5s
-      retries: 24
-      start_period: 10s
-  litellm:
-    image: docker.litellm.ai/berriai/litellm:main-latest
-    depends_on:
-      db:
-        condition: service_healthy
-      langfuse-web:
-        condition: service_healthy
-    environment:
-      - DATABASE_URL=postgresql://${LITELLM_DB_USER}:${LITELLM_DB_PASSWORD}@db:5432/${LITELLM_DB_NAME}
-      - LANGFUSE_PUBLIC_KEY=${LANGFUSE_INIT_PROJECT_PUBLIC_KEY}
-      - LANGFUSE_SECRET_KEY=${LANGFUSE_INIT_PROJECT_SECRET_KEY}
-      - LANGFUSE_OTEL_HOST=http://langfuse-web:3000
-      - LANGFUSE_TRACING_ENVIRONMENT=${LANGFUSE_TRACING_ENVIRONMENT}
-      - LITELLM_MASTER_KEY=${LITELLM_MASTER_KEY}
-    extra_hosts:
-      - host.docker.internal:host-gateway
-    secrets:
-      - source: openai_compatible_api_key
-        target: openai_compatible_api_key
-    volumes:
-      - ./litellm:/app/litellm-config:ro
-    entrypoint:
-      - /bin/sh
-      - /app/litellm-config/start-litellm.sh
-    command:
-      - --config
-      - /app/litellm-config/config.yaml
-      - --log_config
-      - /app/litellm-config/logging.json
-      - --port
-      - "4000"
-    healthcheck:
-      test: ["CMD", "python3", "-c", "import sys, urllib.request; sys.exit(0 if urllib.request.urlopen('http://127.0.0.1:4000/health/readiness', timeout=5).getcode() == 200 else 1)"]
-      interval: 5s
-      timeout: 5s
-      retries: 24
-      start_period: 20s
-  litellm-ui-proxy:
-    build:
-      context: .
-      dockerfile: litellm-ui-proxy/Dockerfile
-    depends_on:
-      litellm:
-        condition: service_healthy
-    environment:
-      - PORT=4000
-      - LITELLM_UPSTREAM_ORIGIN=http://litellm:4000
-      - LITELLM_MASTER_KEY=${LITELLM_MASTER_KEY}
-    healthcheck:
-      test: ["CMD", "bun", "-e", "fetch('http://127.0.0.1:4000/readyz').then((r) => process.exit(r.ok ? 0 : 1)).catch(() => process.exit(1))"]
-      interval: 5s
-      timeout: 5s
-      retries: 24
-      start_period: 10s
-  litellm-bootstrap:
-    image: curlimages/curl:8.16.0
-    depends_on:
-      litellm:
-        condition: service_healthy
-    environment:
-      - LITELLM_MASTER_KEY=${LITELLM_MASTER_KEY}
-      - LITELLM_CLIENT_KEY=${LITELLM_CLIENT_KEY}
-    volumes:
-      - ./litellm/bootstrap-client-key.sh:/bootstrap-client-key.sh:ro
-    entrypoint:
-      - sh
-      - /bootstrap-client-key.sh
-  caddy:
-    image: caddy:2-alpine
-    depends_on:
-      langfuse-proxy:
-        condition: service_healthy
-      litellm:
-        condition: service_healthy
-      litellm-ui-proxy:
-        condition: service_healthy
-    ports:
-      - 127.0.0.1:${TRAICEBOX_PORT}:${TRAICEBOX_PORT}
-    environment:
-      - TRAICEBOX_PORT=${TRAICEBOX_PORT}
-    volumes:
-      - ./caddy/Caddyfile:/etc/caddy/Caddyfile:ro
-    healthcheck:
-      test: ["CMD-SHELL", "wget -qO- --header='Host: litellm.localhost:${TRAICEBOX_PORT}' http://127.0.0.1:${TRAICEBOX_PORT}/health/readiness >/dev/null && wget -qO- --header='Host: litellm.localhost:${TRAICEBOX_PORT}' http://127.0.0.1:${TRAICEBOX_PORT}/ui/ >/dev/null && wget -qO- --header='Host: langfuse.localhost:${TRAICEBOX_PORT}' http://127.0.0.1:${TRAICEBOX_PORT}/readyz >/dev/null"]
-      interval: 5s
-      timeout: 5s
-      retries: 24
-      start_period: 10s
-volumes:
-  postgres_data:
-    driver: local
-  langfuse_clickhouse_data:
-    driver: local
-  langfuse_clickhouse_logs:
-    driver: local
-  langfuse_minio_data:
-    driver: local
-  langfuse_redis_data:
-    driver: local
-secrets:
-  openai_compatible_api_key:
-    file: ${OPENAI_COMPATIBLE_API_KEY_SECRET_FILE:-/dev/null}