How assyaeljou/autonomous-ai-testing-agent works
Autonomous AI Testing Agent is a human-approved web-testing platform for authorized target sites. Users create a session, explore the site for browser evidence, generate and deterministically validate scenarios, review and approve them, execute them through a constrained action registry, then classify failures and produce reports. QA derives read-only campaign, risk, coverage, and evidence-gap views from persisted results. The graph follows documented workflow and composition boundaries; unsampled subsystems are included from README capabilities, but their internal calls are not inferred.
Main entry points: workflow.py LangGraph workflow · repositories.py Persistence repositories · explorer.py Explorer agent · app.py FastAPI
Components 22
User and API
- Streamlit UI
src/ai_testing_agent/frontend/app.py - FastAPI
src/ai_testing_agent/api/app.py - Session service
src/ai_testing_agent/sessions/service.py
Jobs and workflow
- Job runner
src/ai_testing_agent/jobs/runner.py - LangGraph workflow
src/ai_testing_agent/graph/workflow.py
Exploration and security
- Explorer agent
src/ai_testing_agent/exploration/explorer.py - Browser controls
src/ai_testing_agent/browser/action_registry.py - URL scope guard
src/ai_testing_agent/security/url_validator.py
Scenario design
- Test designer
src/ai_testing_agent/test_generation/designer.py - Scenario validator
src/ai_testing_agent/test_generation/validator.py - LLM client
src/ai_testing_agent/llm/client.py - Scenario approval
src/ai_testing_agent/api/routers/scenarios.py
Execution and insights
- Test executor
src/ai_testing_agent/execution/executor.py - Pytest runtime
src/ai_testing_agent/execution/runtime_runner.py - Failure analysis
src/ai_testing_agent/analysis/service.py - QA coverage
src/ai_testing_agent/qa/analysis.py - Report generator
src/ai_testing_agent/reporting/reporter.py - SQLite database
src/ai_testing_agent/storage/db.py - Persistence repositories
src/ai_testing_agent/storage/repositories.py
Other components
- Human tester
- Authorized website
- LLM provider
Connections 28
- Human tester → Streamlit UI: uses
- Streamlit UI → FastAPI: HTTP requests
- FastAPI → Job runner: dispatches jobs
- Job runner → LangGraph workflow: runs stages
src/ai_testing_agent/jobs/runner.py - LangGraph workflow → Explorer agent: invokes exploration
- Explorer agent → Authorized website: observes pages
- Explorer agent → Browser controls: uses controls
- Explorer agent → URL scope guard: checks navigation
src/ai_testing_agent/exploration/explorer.py - LangGraph workflow → Test designer: invokes generation
- Test designer → LLM client: requests drafts
src/ai_testing_agent/test_generation/service.py - LLM client → LLM provider: structured requests
- Test designer → Scenario validator: validates drafts
src/ai_testing_agent/test_generation/service.py - Human tester → Scenario approval: reviews scenarios
- Scenario approval → Test executor: approved scenarios
- LangGraph workflow → Test executor: invokes execution
- Test executor → Pytest runtime: runs bundle
- Pytest runtime → Browser controls: dispatches actions
- Pytest runtime → Authorized website: tests pages
- LangGraph workflow → Failure analysis: invokes analysis
- Failure analysis → LLM client: optional hypothesis
- Failure analysis → Persistence repositories: reads execution rows
src/ai_testing_agent/analysis/service.py - LangGraph workflow → Report generator: generates reports
- Report generator → Persistence repositories: reads persisted rows
src/ai_testing_agent/reporting/reporter.py - FastAPI → QA coverage: serves QA views
- QA coverage → Persistence repositories: reads stored evidence
src/ai_testing_agent/qa/service.py - Session service → Persistence repositories: persists sessions
src/ai_testing_agent/sessions/service.py - Persistence repositories → SQLite database: queries and writes
- Job runner → Persistence repositories: persists jobs
src/ai_testing_agent/jobs/runner.py