Add AML investigation agent

[fcogidi]

Summary

Adds an AML Investigation workflow (case generation + agent runner) and a read-only SQL query tool with strong safety checks, plus a focused test suite for the new functionality.

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 💥 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📝 Documentation update
• 🔧 Refactoring (no functional changes)
• ⚡ Performance improvement
• 🧪 Test improvements
• 🔒 Security fix

Changes Made

• Added AML case parsing/generation utilities (aieng-eval-agents/aieng/agent_evals/aml_investigation/data/cases.py) and shared dataset helpers (aieng-eval-agents/aieng/agent_evals/aml_investigation/data/utils.py).
• Added a read-only SQL database tool with an explicit policy object and safety enforcement (aieng-eval-agents/aieng/agent_evals/tools/sql_database.py) and comprehensive tests (aieng-eval-agents/tests/aieng/agent_evals/tools/test_sql_database.py).
• Added an AML investigation implementation with an ADK-discoverable root_agent, lazy DB initialization, and a CLI for dataset/db/case generation (implementations/aml_investigation/agent.py, implementations/aml_investigation/data/cli.py, implementations/aml_investigation/data/schema.ddl).
• Updated configuration plumbing to support nested env vars (e.g., AML_DB__...) and Langfuse base URL naming (aieng-eval-agents/aieng/agent_evals/configs.py, .env.example).
• Added tests for AML cases + utilities (aieng-eval-agents/tests/aieng/agent_evals/aml_investigation/data/test_cases.py, aieng-eval-agents/tests/aieng/agent_evals/aml_investigation/data/test_utils.py) and updated .gitignore to ignore .adk and *.db.

Testing

• Tests pass locally (uv run pytest tests/)
• Type checking passes (uv run mypy <src_dir>)
• Linting passes (uv run ruff check src_dir/)
• Manual testing performed (describe below)

Manual testing details:

• Ran targeted pytest suites:
  • pytest aieng-eval-agents/tests/aieng/agent_evals/tools/test_sql_database.py
  • pytest aieng-eval-agents/tests/aieng/agent_evals/aml_investigation/data/test_cases.py
  • pytest aieng-eval-agents/tests/aieng/agent_evals/aml_investigation/data/test_utils.py

Screenshots/Recordings

N/A

Related Issues

N/A

Deployment Notes

• New env vars:
  • AML_DB__DRIVER, AML_DB__DATABASE, AML_DB__QUERY__MODE (see .env.example)
  • LANGFUSE_BASE_URL replaces LANGFUSE_HOST in config parsing

Checklist

• Code follows the project's style guidelines
• Self-review of code completed
• Documentation updated (if applicable)
• No sensitive information (API keys, credentials) exposed

[Copilot]

Pull request overview

This PR adds an AML Investigation workflow that enables agents to analyze financial transactions for money laundering patterns. The implementation includes database tooling with strong read-only safety enforcement, case generation utilities, and a complete agent runner with evaluation metrics.

Changes:

• Added AML case generation and dataset utilities with Kaggle integration
• Implemented a read-only SQL database tool with AST-based query validation
• Created an AML investigation agent using Google ADK with database query capabilities

Reviewed changes

Copilot reviewed 20 out of 24 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
pyproject.toml	Removed unused dependencies and added google-adk
implementations/aml_investigation/data/schema.ddl	Database schema for accounts, transactions, and views
implementations/aml_investigation/data/cli.py	CLI commands for database and case file generation
implementations/aml_investigation/agent.py	AML investigation agent implementation with ADK
implementations/aml_investigation/README.md	Template documentation for the implementation
aieng-eval-agents/tests/aieng/agent_evals/tools/test_sql_database.py	Comprehensive tests for SQL database tool
aieng-eval-agents/tests/aieng/agent_evals/aml_investigation/data/test_utils.py	Tests for AML dataset utilities
aieng-eval-agents/tests/aieng/agent_evals/aml_investigation/data/test_cases.py	Tests for case generation and parsing
aieng-eval-agents/pyproject.toml	Added sqlalchemy and sqlglot dependencies
aieng-eval-agents/aieng/agent_evals/tools/sql_database.py	Read-only SQL database tool implementation
aieng-eval-agents/aieng/agent_evals/tools/init.py	Tools module initialization
aieng-eval-agents/aieng/agent_evals/knowledge_agent/tracing.py	Updated to use SecretStr for credentials
aieng-eval-agents/aieng/agent_evals/configs.py	Added database config and SecretStr for sensitive fields
aieng-eval-agents/aieng/agent_evals/async_utils.py	Async utilities for rate limiting and progress tracking
aieng-eval-agents/aieng/agent_evals/async_client_manager.py	Updated to handle SecretStr fields
aieng-eval-agents/aieng/agent_evals/aml_investigation/data/utils.py	Dataset download and normalization utilities
aieng-eval-agents/aieng/agent_evals/aml_investigation/data/cases.py	Case generation and parsing logic
aieng-eval-agents/aieng/agent_evals/aml_investigation/data/init.py	Data module initialization
aieng-eval-agents/aieng/agent_evals/aml_investigation/init.py	AML investigation module initialization
.env.example	Added AML database configuration examples

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.