Merge pull request #5 from bishalprasad321/groq-llm-support

Groq llm support

Author: bishalprasad321

.github/workflows/action-test.yml

@@ -37,7 +37,7 @@ jobs:
   # ============================================================================
   # First, generate an AI-powered PR description using prpilot-summary.
   # This provides context for the multi-model consensus review.
-  # Requires secrets: GEMINI_API_KEY
+  # Requires secrets: GROQ_API_KEY
   # ============================================================================
   generate-description:
     name: "Step 1: Generate PR Description"
@@ -52,10 +52,10 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v4
 
-      - name: Check if GEMINI_API_KEY exists
+      - name: Check if LLM API key exists
         id: check_secrets
         run: |
-          if [ -z "${{ secrets.GEMINI_API_KEY }}" ]; then
+          if [ -z "${{ secrets.GROQ_API_KEY }}" ]; then
             echo "has_key=false" >> $GITHUB_OUTPUT
           else
             echo "has_key=true" >> $GITHUB_OUTPUT
@@ -67,9 +67,9 @@ jobs:
         uses: bishalprasad321/prpilot-summary@v1
         with:
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          llm_api_key: ${{ secrets.GEMINI_API_KEY }}
-          llm_provider: gemini
-          ai_model: gemini-2.5-flash
+          llm_api_key: ${{ secrets.GROQ_API_KEY }}
+          llm_provider: groq
+          ai_model: groq-1.5-mini
           max_diff_lines: "5000"
           enable_incremental_diff_processing: "true"
           debug: ${{ inputs.debug_enabled || 'false' }}
@@ -89,7 +89,7 @@ jobs:
           else
             echo "⚠️ Description generation encountered an error"
             echo "This may be expected if:"
-            echo "  - Secrets are not configured (GEMINI_API_KEY missing)"
+            echo "  - Secrets are not configured (GROQ_API_KEY missing)"
             echo "  - No meaningful changes in the PR"
             echo "  - LLM API is temporarily unavailable"
           fi
@@ -100,7 +100,7 @@ jobs:
   # ============================================================================
   # Run multi-model consensus review after description is generated.
   # This calls 3 independent reviewer models + 1 judge model for consensus.
-  # Requires secrets: GEMINI_API_KEY
+  # Requires secrets: GROQ_API_KEY
   # Outputs: Decision (APPROVE/REQUEST_CHANGES/COMMENT), findings, round count
   # ============================================================================
   consensus-review:
@@ -123,31 +123,31 @@ jobs:
           PRESET="${{ inputs.model_preset || 'default' }}"
           case $PRESET in
             high-capability)
-              echo "reviewer_models=gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite" >> $GITHUB_OUTPUT
-              echo "judge_model=gemini-2.5-pro" >> $GITHUB_OUTPUT
-              echo "preset_name=High Capability (Best Quality)" >> $GITHUB_OUTPUT
+              echo "reviewer_models=llama-3.3-70b-versatile,openai/gpt-oss-120b,openai/gpt-oss-20b" >> $GITHUB_OUTPUT
+              echo "judge_model=openai/gpt-oss-120b" >> $GITHUB_OUTPUT
+              echo "preset_name=Groq Maximum Quality" >> $GITHUB_OUTPUT
               ;;
             cost-optimized)
-              echo "reviewer_models=gemini-2.5-flash,gemini-2.5-flash-lite,gemini-2.5-flash-lite" >> $GITHUB_OUTPUT
-              echo "judge_model=gemini-2.5-flash" >> $GITHUB_OUTPUT
-              echo "preset_name=Cost Optimized (Lower Cost)" >> $GITHUB_OUTPUT
+              echo "reviewer_models=llama-3.1-8b-instant,openai/gpt-oss-20b,openai/gpt-oss-20b" >> $GITHUB_OUTPUT
+              echo "judge_model=openai/gpt-oss-120b" >> $GITHUB_OUTPUT
+              echo "preset_name=Groq Maximum Speed" >> $GITHUB_OUTPUT
               ;;
             quality-optimized)
-              echo "reviewer_models=gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite" >> $GITHUB_OUTPUT
-              echo "judge_model=gemini-2.5-pro" >> $GITHUB_OUTPUT
-              echo "preset_name=Quality Optimized (Best Findings)" >> $GITHUB_OUTPUT
+              echo "reviewer_models=llama-3.3-70b-versatile,openai/gpt-oss-120b,openai/gpt-oss-20b" >> $GITHUB_OUTPUT
+              echo "judge_model=openai/gpt-oss-120b" >> $GITHUB_OUTPUT
+              echo "preset_name=Groq Maximum Quality" >> $GITHUB_OUTPUT
               ;;
             *)
-              echo "reviewer_models=gemini-2.5-flash,gemini-2.5-flash-lite,gemini-2.5-pro" >> $GITHUB_OUTPUT
-              echo "judge_model=gemini-2.5-pro" >> $GITHUB_OUTPUT
-              echo "preset_name=Default (Balanced)" >> $GITHUB_OUTPUT
+              echo "reviewer_models=llama-3.1-8b-instant,openai/gpt-oss-20b,llama-3.3-70b-versatile" >> $GITHUB_OUTPUT
+              echo "judge_model=openai/gpt-oss-120b" >> $GITHUB_OUTPUT
+              echo "preset_name=Groq Default Production" >> $GITHUB_OUTPUT
               ;;
           esac
 
-      - name: Check if GEMINI_API_KEY exists
+      - name: Check if LLM API key exists
         id: check_secrets
         run: |
-          if [ -z "${{ secrets.GEMINI_API_KEY }}" ]; then
+          if [ -z "${{ secrets.GROQ_API_KEY }}" ]; then
             echo "has_key=false" >> $GITHUB_OUTPUT
           else
             echo "has_key=true" >> $GITHUB_OUTPUT
@@ -159,7 +159,9 @@ jobs:
         uses: ./
         with:
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          gemini_api_key: ${{ secrets.GEMINI_API_KEY }}
+          llm_provider: groq
+          llm_api_key: ${{ secrets.GROQ_API_KEY }}
+          llm_provider_url: https://api.groq.com/openai/v1
           reviewer_models: ${{ steps.model_config.outputs.reviewer_models }}
           judge_model: ${{ steps.model_config.outputs.judge_model }}
           max_consensus_rounds: "3"
@@ -198,8 +200,8 @@ jobs:
           else
             echo "⚠️ Consensus review encountered an error"
             echo "This may be expected if:"
-            echo "  - Secrets are not configured (GEMINI_API_KEY missing)"
-            echo "  - Selected models are not valid Gemini v1beta model codes"
+            echo "  - Secrets are not configured (GROQ_API_KEY missing)"
+            echo "  - Selected models are not valid for the current provider"
             echo "  - API quota exceeded (check API key billing settings)"
             echo "  - Reviewer/judge models not available"
             echo "  - LLM API is temporarily unavailable"

README.md

@@ -1,13 +1,13 @@
 # PR Pilot Review
 
-> **Intelligent Multi-Model Consensus AI PR Reviewer** — Automated code reviews with consensus from multiple Gemini AI models.
+> **Intelligent Multi-Model Consensus AI PR Reviewer** — Automated code reviews with consensus from configurable LLM providers like Gemini and Groq.
 
 [![Build Status](https://github.com/bishalprasad321/prpilot-review/actions/workflows/ci.yml/badge.svg)](https://github.com/bishalprasad321/prpilot-review/actions)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 
 ## Overview
 
-PR Pilot Review is a GitHub Action that performs intelligent, consensus-based code reviews on pull requests using Google's Gemini AI models. Unlike single-model reviewers, it runs **3 independent AI reviewers in parallel** and uses a **judge model to reach consensus**, ensuring more thorough and balanced code quality feedback.
+PR Pilot Review is a GitHub Action that performs intelligent, consensus-based code reviews on pull requests using configurable LLM providers such as Gemini or Groq. Unlike single-model reviewers, it runs **3 independent AI reviewers in parallel** and uses a **judge model to reach consensus**, ensuring more thorough and balanced code quality feedback.
 
 ### Key Features
 
@@ -22,20 +22,28 @@ PR Pilot Review is a GitHub Action that performs intelligent, consensus-based co
 
 ## Quick Start
 
-### 1. Get a Gemini API Key
+### 1. Choose your LLM provider
 
-1. Visit [Google AI Studio](https://makersuite.google.com/app/apikey)
-2. Click "Create API Key"
-3. Copy the key
+This action supports multiple providers today, including `gemini` and `groq`.
 
-### 2. Add to GitHub Secrets
+### 2. Add the API key to GitHub Secrets
 
 Go to your repository:
 
 - **Settings** → **Secrets and variables** → **Actions**
-- Add secret: `GEMINI_API_KEY` (paste your key)
+- Add a provider-specific secret, e.g. `GROQ_API_KEY`
 
-### 3. Use the Action
+### 3. Optional base URL for Groq OpenAI-compatible API
+
+If you are using Groq's OpenAI-compatible models, set `llm_provider_url` to:
+
+```yaml
+llm_provider_url: https://api.groq.com/openai/v1
+```
+
+This will route requests to `https://api.groq.com/openai/v1/chat/completions` for chat-based model inference.
+
+### 4. Use the Action
 
 Create `.github/workflows/ai-review.yml`:
 
@@ -61,7 +69,9 @@ jobs:
       - uses: bishalprasad321/prpilot-review@v1
         with:
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          gemini_api_key: ${{ secrets.GEMINI_API_KEY }}
+          llm_provider: groq
+          llm_api_key: ${{ secrets.GROQ_API_KEY }}
+          llm_provider_url: https://api.groq.com/openai/v1
           # Optional: customize model configuration
           # reviewer_models: "gemini-2.5-flash,gemini-2.5-flash-lite,gemini-2.5-pro"
           # judge_model: "gemini-2.5-pro"
@@ -79,7 +89,9 @@ All inputs are optional (sensible defaults provided):
 with:
   # Required
   github_token: ${{ secrets.GITHUB_TOKEN }}
-  gemini_api_key: ${{ secrets.GEMINI_API_KEY }}
+  llm_provider: groq
+  llm_api_key: ${{ secrets.GROQ_API_KEY }}
+  llm_provider_url: https://api.groq.com/openai/v1
 
   # Optional: Model Configuration
   reviewer_models: "gemini-2.5-flash,gemini-2.5-flash-lite,gemini-2.5-pro" # 3 models
@@ -93,47 +105,51 @@ with:
   debug: "false" # Verbose logging
 ```
 
-Use stable Gemini `v1beta` model codes. If you still pass legacy aliases such as `gemini-3.0-flash` or `gemini-3.1-pro`, the action will try to remap them to supported 2.x models before calling the API.
+Use stable provider-specific model IDs. For Gemini, use `v1beta` model codes. For Groq, use models like `llama-3.1-8b-instant`, `openai/gpt-oss-20b`, or `llama-3.3-70b-versatile` (check [Groq console](https://console.groq.com/keys) for available models).
 
-### Model Presets
-
-Choose models based on your needs:
-
-#### Default (Recommended) ✅
+### Groq Example
 
 ```yaml
-reviewer_models: "gemini-2.5-flash,gemini-2.5-flash-lite,gemini-2.5-pro"
-judge_model: "gemini-2.5-pro"
+with:
+  github_token: ${{ secrets.GITHUB_TOKEN }}
+  llm_provider: groq
+  llm_api_key: ${{ secrets.GROQ_API_KEY }}
+  reviewer_models: "llama-3.1-8b-instant,openai/gpt-oss-20b,llama-3.3-70b-versatile"
+  judge_model: "openai/gpt-oss-120b"
+  max_consensus_rounds: "3"
+  inline_comments_enabled: "true"
 ```
 
-Best balance of speed, quality, and cost.
+### Model Presets
+
+Choose models based on your needs:
 
-#### High Capability (Best Quality)
+#### Groq Default Production ✅
 
 ```yaml
-reviewer_models: "gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite"
-judge_model: "gemini-2.5-pro"
+reviewer_models: "llama-3.1-8b-instant,openai/gpt-oss-20b,llama-3.3-70b-versatile"
+judge_model: "openai/gpt-oss-120b"
 ```
 
-Uses most capable models. Higher cost but best findings.
+Best balance of production reliability and coverage.
 
-#### Cost Optimized (Lowest Cost)
+#### Groq Maximum Speed ✅
 
 ```yaml
-reviewer_models: "gemini-2.5-flash,gemini-2.5-flash-lite,gemini-2.5-flash-lite"
-judge_model: "gemini-2.5-flash"
+reviewer_models: "llama-3.1-8b-instant,openai/gpt-oss-20b,openai/gpt-oss-20b"
+judge_model: "openai/gpt-oss-120b"
 ```
 
-Uses lighter models. Lower cost but lighter analysis.
+Optimized for faster review cycles with lighter reviewer models.
 
-#### Quality Optimized (Most Findings)
+#### Groq Maximum Quality ✅
 
 ```yaml
-reviewer_models: "gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite"
-judge_model: "gemini-2.5-pro"
+reviewer_models: "llama-3.3-70b-versatile,openai/gpt-oss-120b,openai/gpt-oss-20b"
+judge_model: "openai/gpt-oss-120b"
 ```
 
-Maximum detection power. Highest cost, most comprehensive reviews.
+Highest quality review configuration for the deepest analysis.
 
 ## How It Works
 
@@ -194,7 +210,7 @@ The action posts a GitHub review with:
 See detailed architecture in [docs/adr/](./docs/adr/):
 
 - **ADR-001** — Multi-Model Consensus Architecture
-- **ADR-002** — Gemini API Integration & Model Selection
+- **ADR-002** — LLM Provider Integration & Model Selection
 - **ADR-003** — Inline Comment Mapping
 - **ADR-004** — Idempotency & State Management
 
@@ -205,7 +221,7 @@ prpilot-review/
 ├── src/
 │   ├── index.ts                 # Main orchestrator (13-step pipeline)
 │   ├── llm/
-│   │   └── llm-client.ts       # Gemini API abstraction
+│   │   └── llm-client.ts       # LLM provider abstraction
 │   ├── review/
 │   │   ├── review-orchestrator.ts  # Consensus logic
 │   │   └── inline-comment-builder.ts   # Line mapping
@@ -285,12 +301,11 @@ npm run all          # Run all checks
 
 ## API Keys & Billing
 
-### Getting a Gemini API Key
+### Getting an API Key
 
-1. Go to [Google AI Studio](https://makersuite.google.com/app/apikey)
-2. Click "Create API Key"
-3. Select/create a project
-4. Copy the key and add to GitHub repo secrets
+1. Choose your provider: `gemini` or `groq`
+2. Create an API key in the provider console
+3. Copy the key and add it to GitHub repo secrets
 
 ### Free Tier Quotas
 
@@ -302,48 +317,48 @@ The free tier has limits:
 If you hit quota errors:
 
 1. **Upgrade to paid plan** — Recommended for production
-2. **Use lighter models** — `gemini-2.5-flash-lite` uses fewer tokens
+2. **Use lighter models** — `gemini-2.5-flash-lite` or Groq's smaller models like `llama-3.1-8b-instant` use fewer tokens
 3. **Reduce reviewer count** — Use 2 instead of 3 reviewers
 4. **Schedule reviews** — Spread runs across off-peak hours
 
-See [quota documentation](https://ai.google.dev/gemini-api/docs/rate-limits).
+See provider-specific quota documentation for your chosen LLM.
 
 ### Costs
 
-Typical cost per PR review:
+Typical cost per PR review will vary by provider and model selection:
 
-- **Default config** — $0.001-0.005 per review (3 reviewers + optional judge)
-- **Cost-optimized** — $0.0005-0.002 per review
-- **Quality-optimized** — $0.005-0.02 per review
+- **Default config** — Low-cost balanced review
+- **Cost-optimized** — Minimal token usage and fastest reviewers
+- **Quality-optimized** — More comprehensive analysis with higher cost
 
-Pricing based on [Gemini API rates](https://ai.google.dev/pricing).
+Pricing is provider-dependent. For Gemini, see [Gemini pricing](https://ai.google.dev/pricing).
 
 ## Troubleshooting
 
 ### Action not running?
 
-- Check `GEMINI_API_KEY` is set in repo secrets
+- Check `llm_provider` and `llm_api_key` are configured correctly
 - Verify branch triggers (pull_request, branches)
 - Check action logs for errors
 
 ### Zero findings even with code changes?
 
 - Enable debug mode to see full model responses
-- Check if models are available in v1beta API
-- Try different model preset
-- Verify PR has actual code changes
+- Check whether the selected models are valid for your provider
+- Try a different model preset
+- Verify the PR has actual code changes
 
 ### Quota exceeded errors?
 
-- Check your billing plan at [Google Cloud Console](https://console.cloud.google.com)
-- Switch to paid plan or upgrade quota
-- Use lighter models (flash-lite variants)
+- Check your provider billing or free tier quota
+- Switch to paid plan or upgrade quota if available
+- Use lighter models where supported
 
 ### Model not found errors?
 
-- Verify model names match available list
-- Models must be in `v1beta` API
-- Check [available models](https://ai.google.dev/gemini-api/docs/models/gemini)
+- Verify model names match your provider's available models
+- For Gemini, check [Gemini models](https://ai.google.dev/gemini-api/docs/models/gemini)
+- For Groq, check your Groq account model docs
 
 ### Slow reviews?
 
@@ -439,7 +454,7 @@ MIT License — See [LICENSE](./LICENSE) file
 
 ## Acknowledgments
 
-- Built with [Gemini AI API](https://ai.google.dev/)
+- Built with configurable LLM providers such as Gemini or Groq
 - GitHub Actions integration via [@actions/core](https://github.com/actions/toolkit)
 - Code bundling via [@vercel/ncc](https://github.com/vercel/ncc)