Summary
Add more tests to cover plugin failure modes in Open Agentic 2.0.
The goal is to make sure the orchestrator behaves correctly when plugins fail
(e.g. timeouts, bad JSON, or missing evidence).
What to test
Where possible, add or extend tests in tests/test_agentic2.py (or a new test
file if that makes more sense) to cover cases like:
- a plugin call that times out (e.g. subprocess / HTTP timeout),
- a plugin that returns invalid JSON or non-JSON output,
- a plugin that returns a response without evidence or with too-low coverage.
When this is done
- There are automated tests that simulate plugin failures (timeout / bad JSON /
missing or low-quality evidence).
- The orchestrator:
- logs these failures in the audit chain, and
- either blocks / abstains as expected (fail-closed), instead of silently
returning ok=True.
- Running
pytest -q -vv still passes.
Summary
Add more tests to cover plugin failure modes in Open Agentic 2.0.
The goal is to make sure the orchestrator behaves correctly when plugins fail
(e.g. timeouts, bad JSON, or missing evidence).
What to test
Where possible, add or extend tests in
tests/test_agentic2.py(or a new testfile if that makes more sense) to cover cases like:
When this is done
missing or low-quality evidence).
returning
ok=True.pytest -q -vvstill passes.