Running Tests¶

Overview¶

All ADEPT tests are invoked through Makefile targets. This ensures consistent execution environments, proper network configuration, and reproducible results across development machines and CI.

Prerequisites¶

Before running tests, ensure:

1. Services are running -- execute make start to bring up the full stack.
2. Credentials are synced -- run make sync-credentials to copy service secrets to the test credential directory.
3. Docker is available -- all tests execute inside containers.

# Verify prerequisites
make start
make sync-credentials
docker ps --format "table {{.Names}}\t{{.Status}}" | head -10

Quick Reference¶

Running Specific Tests¶

To run a single test file directly:

docker run --rm \
  --network=adept_application_network \
  --network=adept_frontend_network \
  -v "$(pwd)/src":/app/src:ro \
  -v "$(pwd)/tests":/app/tests:ro \
  -e AGENT_GATEWAY_URL=http://agent_gateway:8081/v1 \
  -e KEYCLOAK_URL=http://keycloak:8180 \
  agentic-framework-deps-base \
  /app/.venv/bin/pytest tests/e2e/test_example.py -v -s

Test Output¶

All test invocations should be wrapped with nohup and directed to timestamped log files:

mkdir -p logs
nohup make validate-unit-all > logs/unit_tests_$(date +%Y%m%d_%H%M%S).log 2>&1 &

This pattern ensures:

• Long-running tests do not block your terminal.
• Output is preserved for later analysis.
• Timestamps enable correlation with service logs.

To check results:

# View the most recent log
ls -t logs/ | head -1 | xargs -I{} tail -30 logs/{}

# Search for failures
grep -l "FAILED" logs/*.log

Filtering Noisy Output¶

Some targets produce verbose pip installation messages. Filter them for cleaner output:

make validate-response-api-tool-calls 2>&1 | grep -v "Requirement already satisfied" | tail -40

Troubleshooting¶