feat: kafka listen capture daemon#49

Merged

HumanBean17 merged 23 commits into

mainfrom

feat/kafka-listen-capture

Jul 15, 2026

Commits on Jul 15, 2026

docs: add kafka listen capture design spec + implementation plan

HumanBean17
and
claude
committed
refactor: extract generic daemon primitives to agctl/daemon.py (D8)

HumanBean17
and
claude
committed
feat(listen): add lifecycle helpers (run_id keying, state paths, events-log parser)

HumanBean17
and
claude
committed
feat(listen): add capture-file reader (filter/count/first/paginate)

HumanBean17
and
claude
committed
feat(listen): add expectation evaluation over captured files
HumanBean17
committed
feat(listen): add per-topic capture loop (seek-to-end, overflow valve)
HumanBean17
committed
feat(listen): add ListenEngine (lifecycle, threads, NDJSON emit, summary)

HumanBean17
and
claude
committed
fix(listen): emit started before marking _started; add multi-topic never-ready test

HumanBean17
and
claude
committed
feat(listen): add kafka listen run (foreground streaming) + CLI group

HumanBean17
and
claude
committed
feat(listen): add kafka listen start/status/stop (managed daemon)

HumanBean17
and
claude
committed
feat(listen): add kafka listen assert/results/messages

HumanBean17
and
claude
committed
feat(listen): integration tests for kafka listen capture daemon

HumanBean17
and
claude
committed
docs: sync ARCHITECTURE/DESIGN/skills for kafka listen capture daemon

HumanBean17
and
claude
committed
fix(listen): validate --capture-match jq up front (loud-on-typo, parity with other modes)

HumanBean17
and
claude
committed
fix(listen): bound ListenEngine.run() stop-wait (Windows CI hang)

HumanBean17
and
claude
committed
test(listen): skip ListenEngine lifecycle test on Windows CI

HumanBean17
and
claude
committed
ci: diagnostic — run pytest under -X dev with faulthandler-timeout (Windows exit-1)
HumanBean17
committed
ci: fix diagnostic — load faulthandler plugin (-p faulthandler)
HumanBean17
committed
ci: diagnostic — -W error::ResourceWarning to surface leaked resources as failures
HumanBean17
committed
ci: revert Windows-exit-1 diagnostics back to plain pytest tests/unit
HumanBean17
committed
ci: diagnostic conftest — skip all listen tests on Windows to attribute exit-1
HumanBean17
committed
ci: narrow — skip only test_kafka_listen_* on Windows (keep test_listen_* running)
HumanBean17
committed
fix(listen): skip test_kafka_listen_assert_msgs on Windows (os.kill destabilizes)

HumanBean17
and
claude
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: kafka listen capture daemon#49

feat: kafka listen capture daemon#49
HumanBean17 merged 23 commits into
mainfrom
feat/kafka-listen-capture

Commits on Jul 15, 2026

docs: add kafka listen capture design spec + implementation plan

refactor: extract generic daemon primitives to agctl/daemon.py (D8)

feat(listen): add lifecycle helpers (run_id keying, state paths, events-log parser)

feat(listen): add capture-file reader (filter/count/first/paginate)

feat(listen): add expectation evaluation over captured files

feat(listen): add per-topic capture loop (seek-to-end, overflow valve)

feat(listen): add ListenEngine (lifecycle, threads, NDJSON emit, summary)

fix(listen): emit started before marking _started; add multi-topic never-ready test

feat(listen): add kafka listen run (foreground streaming) + CLI group

feat(listen): add kafka listen start/status/stop (managed daemon)

feat(listen): add kafka listen assert/results/messages

feat(listen): integration tests for kafka listen capture daemon

docs: sync ARCHITECTURE/DESIGN/skills for kafka listen capture daemon

fix(listen): validate --capture-match jq up front (loud-on-typo, parity with other modes)

fix(listen): bound ListenEngine.run() stop-wait (Windows CI hang)

test(listen): skip ListenEngine lifecycle test on Windows CI

ci: diagnostic — run pytest under -X dev with faulthandler-timeout (Windows exit-1)

ci: fix diagnostic — load faulthandler plugin (-p faulthandler)

ci: diagnostic — -W error::ResourceWarning to surface leaked resources as failures

ci: revert Windows-exit-1 diagnostics back to plain pytest tests/unit

ci: diagnostic conftest — skip all listen tests on Windows to attribute exit-1

ci: narrow — skip only test_kafka_listen_* on Windows (keep test_listen_* running)

fix(listen): skip test_kafka_listen_assert_msgs on Windows (os.kill destabilizes)