v0.2.0: Simplify provider, fix compose-openwebui mounts, clean up build

Provider:
- Extract duplicated usage-parsing into _extract_usage()
- Replace if/elif role chain with ROLE_PREFIXES dict lookup
- datetime.now().timestamp() → time.time()
- Remove redundant else branch in completion()
- Rename internal methods to private (_format_prompt, _extract_model, _build_response)

Bug fix:
- compose-openwebui.yaml: wrong volume mount paths (./providers:/app/providers
  instead of ./providers:/app/config/providers, ./config:/app/config instead of
  file-level mount) — would break provider loading at runtime

Build:
- Remove redundant BUILDKIT_INLINE_CACHE=1 (already using GHA cache)
- Consolidate 3 separate docker run test invocations into one
- Clean up release job (inline variables, deduplicate comments)

Cleanup:
- Remove redundant Dockerfile/entrypoint comments
- Remove overly broad .gitignore wildcards (*api*key*, *secret*, *token*)
- Fix .env.openwebui.example reference to nonexistent compose file
- Consolidate CHANGELOG v0.1.0 duplicate sections

Author: cabinlab

.env.example

@@ -1,10 +1,8 @@
 # LiteLLM Environment Configuration
 
 # REQUIRED: Master key protects access to your authenticated Claude instance
-# Keys MUST start with 'sk-' (LiteLLM requirement)
-# Replace with your own key before use
-# Generate a secure key: echo "sk-$(openssl rand -hex 32)"
-# Or for development: sk-dev-test-key
+# Must start with 'sk-' (LiteLLM requirement)
+# Generate one: echo "sk-$(openssl rand -hex 32)"
 LITELLM_MASTER_KEY=sk-your-desired-custom-key
 
 # Database configuration (required for LiteLLM)

.env.openwebui.example

@@ -1,8 +1,5 @@
-# Open WebUI Test Configuration
-# Copy this to .env to use with compose-openwebui.yaml or compose-openwebui-https.yaml
-
-# Use existing LiteLLM configuration from main .env file
-# Or set these values for testing:
+# Open WebUI Configuration
+# Copy this to .env to use with compose-openwebui.yaml
 
 # REQUIRED: Master key for LiteLLM API access
 LITELLM_MASTER_KEY=sk-your-test-key-here

.github/workflows/build-and-publish.yml

@@ -69,8 +69,6 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           cache-from: type=gha
           cache-to: type=gha,mode=max
-          build-args: |
-            BUILDKIT_INLINE_CACHE=1
 
   test:
     if: github.event_name == 'pull_request'
@@ -83,28 +81,17 @@ jobs:
       - name: Build test image
         run: docker build -t litellm-claude-test .
 
-      - name: Test container
+      - name: Verify container dependencies
         run: |
-          # Test that the container can start
           docker run --rm \
             -e LITELLM_MASTER_KEY=sk-test-key \
             -e CLAUDE_CODE_OAUTH_TOKEN=sk-ant-oat01-test \
             litellm-claude-test \
-            python --version
-
-          # Test that LiteLLM is installed
-          docker run --rm \
-            -e LITELLM_MASTER_KEY=sk-test-key \
-            -e CLAUDE_CODE_OAUTH_TOKEN=sk-ant-oat01-test \
-            litellm-claude-test \
-            litellm --version || echo "LiteLLM version check"
-
-          # Test that the Claude Agent SDK is importable
-          docker run --rm \
-            -e LITELLM_MASTER_KEY=sk-test-key \
-            -e CLAUDE_CODE_OAUTH_TOKEN=sk-ant-oat01-test \
-            litellm-claude-test \
-            python -c "import claude_agent_sdk; print('Claude Agent SDK available')"
+            bash -c '
+              python --version &&
+              litellm --version &&
+              python -c "import claude_agent_sdk; print(\"Claude Agent SDK available\")"
+            '
 
   release:
     needs: build
@@ -120,44 +107,33 @@ jobs:
           fetch-depth: 0
           token: ${{ secrets.GITHUB_TOKEN }}
 
-      - name: Get latest tag
+      - name: Compute next patch version
         id: get_latest_tag
         run: |
-          # Get the latest tag, defaulting to v0.0.0 if no tags exist
           LATEST_TAG=$(git describe --tags --abbrev=0 2>/dev/null || echo "v0.0.0")
           echo "latest_tag=$LATEST_TAG" >> $GITHUB_OUTPUT
-          
-          # Extract version numbers
+
           VERSION=${LATEST_TAG#v}
           IFS='.' read -r MAJOR MINOR PATCH <<< "$VERSION"
-          
-          # Increment patch version
-          NEW_PATCH=$((PATCH + 1))
-          NEW_TAG="v${MAJOR}.${MINOR}.${NEW_PATCH}"
-          
+          NEW_TAG="v${MAJOR}.${MINOR}.$((PATCH + 1))"
+
           echo "new_tag=$NEW_TAG" >> $GITHUB_OUTPUT
-          echo "Latest tag: $LATEST_TAG"
-          echo "New tag: $NEW_TAG"
-      
+          echo "Latest: $LATEST_TAG -> Next: $NEW_TAG"
+
       - name: Generate release notes
-        id: release_notes
         run: |
-          # Get commit messages since last tag
           LATEST_TAG=${{ steps.get_latest_tag.outputs.latest_tag }}
-          
+
           echo "## What's Changed" > release_notes.md
           echo "" >> release_notes.md
-          
-          # Get commit messages since last tag
+
           if [ "$LATEST_TAG" == "v0.0.0" ]; then
             git log --pretty=format:"* %s (%h)" >> release_notes.md
           else
             git log ${LATEST_TAG}..HEAD --pretty=format:"* %s (%h)" >> release_notes.md
           fi
-          
-          echo "" >> release_notes.md
-          echo "" >> release_notes.md
-          echo "**Full Changelog**: https://github.com/${{ github.repository }}/compare/${LATEST_TAG}...${{ steps.get_latest_tag.outputs.new_tag }}" >> release_notes.md
+
+          echo -e "\n\n**Full Changelog**: https://github.com/${{ github.repository }}/compare/${LATEST_TAG}...${{ steps.get_latest_tag.outputs.new_tag }}" >> release_notes.md
       
       - name: Create and push tag
         run: |

.gitignore

@@ -96,14 +96,13 @@ Thumbs.db
 tmp/
 temp/
 
-# API keys and secrets (backup protection)
-*api*key*
-*secret*
-*token*
+# Secrets and credentials
 credentials.*
+*.pem.key
+*.key.pem
 
 # OAuth tokens and Claude credentials
-.claude 
+.claude
 .claude.json
 claude.json
 **/claude.json
@@ -113,11 +112,8 @@ claude.json
 .serena/
 certs/
 
-# Test files and directories
+# Test files
 tests/
 test_*.py
 test-*.sh
 test_*.sh
-
-# Dev research
-.archive/docker-compose.override.yml

CHANGELOG.md

@@ -23,22 +23,12 @@ All notable changes to this project will be documented in this file.
 ## [0.1.0] - 2025-07-19
 
 ### Added
-- Initial release of LiteLLM Claude Code Provider
-- Custom LiteLLM provider implementation for Claude Code SDK
-- Support for all Claude models (Opus, Sonnet, Haiku) via OpenAI-compatible API
-- OAuth authentication support for Claude Pro/Max users
-- Docker-based deployment with automated image publishing
+- Custom LiteLLM provider for Claude Code SDK with OpenAI API compatibility
+- Support for all Claude models (Opus, Sonnet, Haiku)
+- OAuth authentication for Claude Pro/Max users (persists across restarts)
+- Docker-based deployment with multi-platform images (amd64/arm64)
 - Web UI for testing at `localhost:4000/ui/`
-
-### Features
-- OpenAI API compatibility for Claude models
-- Auth persists across container restarts
-- Support for both OAuth (Claude subscription) tokens and API keys
-- Multi-platform Docker images (amd64/arm64)
-
-### Notes
-- This is an early release and has not been tested with LiteLLM-compatible clients
-- Streaming support is partial due to Claude Code SDK limitations
+- Streaming response support (partial, due to Claude Code SDK limitations)
 
 [0.2.0]: https://github.com/cabinlab/litellm-claude-code/releases/tag/v0.2.0
 [0.1.0]: https://github.com/cabinlab/litellm-claude-code/releases/tag/v0.1.0

Dockerfile

@@ -1,18 +1,16 @@
 FROM ghcr.io/cabinlab/claude-code-sdk-docker:python
 
-# Switch to root for installations
 USER root
 
-# Install dependencies from requirements.txt
 COPY requirements.txt /tmp/requirements.txt
 RUN pip install --no-cache-dir -r /tmp/requirements.txt && rm /tmp/requirements.txt
 
-# Set up writable cache directories for claude user
+# Prisma cache must be writable by the non-root runtime user
 ENV PRISMA_PYTHON_CACHE_DIR="/home/claude/.cache/prisma-python"
 RUN mkdir -p /home/claude/.cache/prisma-python && \
     chown -R claude:claude /home/claude/.cache
 
-# Generate Prisma client as the non-root runtime user so binary paths resolve under /home/claude.
+# Generate Prisma client as claude so binary paths resolve under /home/claude
 ENV PATH="/opt/venv/bin:$PATH"
 RUN chown -R claude:claude /opt/venv/lib/python3.11/site-packages/prisma && \
     su -s /bin/bash claude -c 'cd /opt/venv/lib/python3.11/site-packages/litellm/proxy && \
@@ -24,27 +22,18 @@ RUN chown -R claude:claude /opt/venv/lib/python3.11/site-packages/prisma && \
       chmod -R u+rwX,go+rX /opt/venv/lib/python3.11/site-packages/litellm/proxy/_experimental/out; \
     fi
 
-# Copy application code with proper ownership.
-# LiteLLM resolves custom_handler paths relative to /app/config.
+# LiteLLM resolves custom_handler paths relative to /app/config
 COPY --chown=claude:claude config/ /app/config/
 COPY --chown=claude:claude providers/ /app/config/providers/
 
-# Copy entrypoint script
 COPY scripts/entrypoint.sh /app/scripts/entrypoint.sh
 RUN chmod +x /app/scripts/entrypoint.sh
 
-# Switch back to claude user
 USER claude
-
-# Working directory
 WORKDIR /app
-
-# Expose LiteLLM port
 EXPOSE 4000
 
-# Health check (avoid curl dependency; include auth for protected endpoint)
 HEALTHCHECK --interval=30s --timeout=10s --retries=3 \
   CMD python3 -c 'import os, urllib.request; req = urllib.request.Request("http://localhost:4000/health/readiness", headers={"Authorization": "Bearer " + os.environ.get("LITELLM_MASTER_KEY", "")}); urllib.request.urlopen(req, timeout=5)' || exit 1
 
-# Single entrypoint
 ENTRYPOINT ["/app/scripts/entrypoint.sh"]

compose-openwebui.yaml

@@ -19,8 +19,8 @@ services:
     expose:
       - "4000"
     volumes:
-      - ./config:/app/config:ro
-      - ./providers:/app/providers:ro
+      - ./config/litellm_config.yaml:/app/config/litellm_config.yaml:ro
+      - ./providers:/app/config/providers:ro
       - claude-auth:/home/claude/.claude
     env_file:
       - .env
@@ -37,12 +37,11 @@ services:
     volumes:
       - open-webui:/app/backend/data
     environment:
+      - ENABLE_OPENAI_API=true
       - OPENAI_API_KEY=${LITELLM_MASTER_KEY}
       - OPENAI_API_BASE_URL=http://litellm:4000/v1
       - ENABLE_OLLAMA_API=false
-      - OLLAMA_BASE_URL=""
       - WEBUI_SECRET_KEY=${WEBUI_SECRET_KEY:-}
-      - ENABLE_OPENAI_API=true
       - WEBUI_SESSION_COOKIE_SECURE=true
       - WEBUI_AUTH_COOKIE_SECURE=true
     env_file: