How vijay-msrit/ai-ain-t-ready-for-us- works
Fixora accepts a GitHub issue or a manual run request, then indexes the repository, classifies the issue, locates relevant code, generates a patch, evaluates it, and can open a pull request. Its React frontend exposes those five phases and displays pipeline results. The README documents the agent workflow; the excerpts directly show webhook dispatch, API pipeline execution, and localizer retrieval/LLM use. Several agent implementations and graph wiring were not sampled, so their documented workflow links cite the README rather than implying source-level calls.
Main entry points: graph.py Agent Orchestrator · indexer.py Repository Indexer · webhook.py Webhook Triggers · llm_client.py LLM Client
Components 24
Ingress and API
- FastAPI App
app/main.py - Webhook Triggers
app/webhook.py - Pipeline API
app/api.py
Resolution Pipeline
- Agent Orchestrator
app/graph.py - Shared State
app/state.py - Repository Indexer
app/agents/indexer.py - Issue Processor
app/agents/issue_processor.py - Bug Localizer
app/agents/localizer.py - Patch Generator
app/agents/patcher.py - Patch Evaluator
app/agents/evaluator.py
Runtime Tools
- Repository Utilities
app/tools/repo_utils.py - Code Chunker
app/tools/chunker.py - Vector Store
app/tools/vector_store.py - LLM Client
app/tools/llm_client.py - Patch Applicator
app/tools/code_applicator.py - GitHub Client
app/tools/github_client.py - Configuration
app/config.py
Frontend
- Web Interface
frontend/src/App.jsx - Pipeline Pages
frontend/src/pages - API Client
frontend/src/api/client.js
Other components
- User
- GitHub
- ChromaDB
- LLM Provider
Connections 28
- User → Web Interface: uses
frontend/src/App.jsx - User → Webhook Triggers: submits issue
README.md - FastAPI App → Webhook Triggers: registers router
app/main.py - FastAPI App → Pipeline API: registers router
app/main.py - Webhook Triggers → Agent Orchestrator: dispatches run
app/webhook.py - Pipeline API → Agent Orchestrator: runs pipeline
app/api.py - Agent Orchestrator → Repository Indexer: runs phase
README.md - Agent Orchestrator → Issue Processor: runs phase
README.md - Agent Orchestrator → Bug Localizer: runs phase
README.md - Agent Orchestrator → Patch Generator: runs phase
README.md - Agent Orchestrator → Patch Evaluator: runs phase
README.md - Repository Indexer → Repository Utilities: clones repository
README.md - Repository Indexer → Code Chunker: chunks code
README.md - Repository Indexer → Vector Store: indexes chunks
README.md - Vector Store → ChromaDB: reads and writes
README.md - Bug Localizer → Vector Store: queries index
app/agents/localizer.py - Bug Localizer → LLM Client: builds and reranks
app/agents/localizer.py - LLM Client → LLM Provider: requests completion
app/tools/llm_client.py - Patch Generator → LLM Client: generates patch
README.md - Patch Generator → Patch Applicator: applies patch
README.md - Patch Evaluator → GitHub Client: opens pull request
README.md - GitHub Client → GitHub: creates branch and PR
app/tools/github_client.py - GitHub → Webhook Triggers: sends issue event
README.md - Web Interface → Pipeline Pages: routes to
frontend/src/App.jsx - Pipeline Pages → API Client: requests pipeline data
frontend/src/pages/Evaluator.jsx - API Client → Pipeline API: calls API
frontend/src/api/client.js - Configuration → LLM Client: configures provider
app/tools/llm_client.py - Configuration → GitHub Client: provides token
app/tools/github_client.py