How ishanjais01io/autodevai works
AutoDevAI clones a GitHub repository, then runs a multi-agent engineering review: planning, code review, testing, documentation, security analysis, and final reporting. The standard workflow and optional CrewAI path are selected by the CLI entry point; agents share repository context through shared memory. A separate dashboard presents scan history, findings, and recommendations, backed by SQLite and analysis helpers. The README documents the agent sequence and report outputs; unsampled modules are included only where the tree or source index supports their role, so some internal calls are not shown.
Main entry points: workflow.py Agent workflow · main.py CLI pipeline · dashboard_service.py Dashboard service · crew.py CrewAI orchestration
Components 30
Orchestration
- CLI pipeline
main.py - Shared memory
core/shared_memory.py - Agent workflow
core/workflow.py - CrewAI orchestration
crew/crew.py
Engineering agents
- Planning agent
agents/planner.py - Review agent
agents/reviewer.py - Review batching
core/review_engine.py - Testing agent
agents/tester.py - Documentation agent
agents/documentation.py - Security agent
agents/security.py - Final report agent
agents/final_report.py - Score calculation
core/score_engine.py - Engineering reports
agents/final_report.py - CrewAI agents
crew/agents.py - CrewAI tasks
crew/tasks.py - Finding analysis
core/finding_engine.py - Recommendations
core/recommendation_engine.py
LLM and tools
- Repository clone
tools/github.py - Security scanner
tools/security_scanner.py - LLM service
services/llm_service.py - Provider selection
llm/provider.py
Dashboard and data
- Dashboard app
dashboard/app.py - Dashboard interface
dashboard/static/dashboard.js - Dashboard API
dashboard/api.py - Dashboard service
dashboard/services/dashboard_service.py - Scan comparison
core/comparison_engine.py - Scan records
storage/models.py - SQLite database
storage/database.py
Other components
- User
- GitHub repository
Connections 30
- User → CLI pipeline: submits URL
README.md - CLI pipeline → Repository clone: clones repository
main.py - Repository clone → GitHub repository: fetches
README.md - CLI pipeline → Shared memory: stores repo path
main.py - CLI pipeline → Agent workflow: runs pipeline
main.py - Agent workflow → Planning agent: runs first
README.md - Agent workflow → Review agent: runs next
README.md - Agent workflow → Testing agent: runs next
README.md - Agent workflow → Documentation agent: runs next
README.md - Agent workflow → Security agent: runs next
README.md - Agent workflow → Final report agent: runs last
README.md - Review agent → Review batching: batches files
agents/reviewer.py - Review agent → LLM service: requests review
agents/reviewer.py - LLM service → Provider selection: delegates review
services/llm_service.py - Security agent → Security scanner: scans repository
agents/security.py - Final report agent → Score calculation: uses scores
agents/final_report.py - Final report agent → Engineering reports: generates reports
README.md - CLI pipeline → SQLite database: initializes
main.py - CLI pipeline → CrewAI orchestration: selects optional path
main.py - CrewAI orchestration → CrewAI agents: builds agents
crew/crew.py - CrewAI orchestration → CrewAI tasks: builds tasks
crew/crew.py - CrewAI agents → Provider selection: uses provider
crew/agents.py - Dashboard app → Dashboard API: mounts routes
dashboard/app.py - Dashboard app → Dashboard interface: serves static files
dashboard/app.py - Dashboard API → Dashboard service: uses dashboard data
dashboard/api.py - Dashboard API → Scan comparison: compares scans
dashboard/api.py - Dashboard service → Scan records: loads latest scan
dashboard/services/dashboard_service.py - Dashboard service → Finding analysis: generates findings
dashboard/services/dashboard_service.py - Dashboard service → Recommendations: generates recommendations
dashboard/services/dashboard_service.py - Scan records → SQLite database: reads and writes
storage/models.py