Self-improving AI Employees for everyone
Python ai-agentsai-employeeai-employeesassistantautonomous-agentsco-employeescompanygrok-botinstinctloop-agentsmuseown-your-datapersonalself-hosted
docs(rfc): record RFC-0005 Phase 1 as landed
trohitg78ebd50c53d5
diff --git a/RFC-0005-EVENT-OS-ON-TEMPORAL.md b/RFC-0005-EVENT-OS-ON-TEMPORAL.mdindex 210febd5..44e7c792 100644--- a/RFC-0005-EVENT-OS-ON-TEMPORAL.md+++ b/RFC-0005-EVENT-OS-ON-TEMPORAL.md@@ -1,6 +1,6 @@# RFC-0005 — Event OS: Event-Driven AI Employees on Temporal-- **Status:** Accepted, 2026-10-23 (owner decisions in §3). Implementation is phased (§24). Phase 0 landed on 2026-10-23 (§24.4); Phase 1 is in progress. RFC-0006 amended this RFC on 2026-10-24 (§30).+- **Status:** Accepted, 2026-10-23 (owner decisions in §3). Implementation is phased (§24). Phase 0 landed on 2026-10-23 and Phase 1 on 2026-10-24 (§24.4); Phase 1b (RFC-0006 §22.2) is next. RFC-0006 amended this RFC on 2026-10-24 (§30).- **Created:** 2026-10-23- **Scope:** the path from every event source to every effect:- `server/services/events/` and `server/routers/webhook.py`;@@ -1145,16 +1145,16 @@ After the upgrade to SDK 1.34.0, new patch ids ship dormant for one release (`pa- **CI:** the whole server suite runs on Linux, Windows and macOS with the pinned CLI.- **Fixtures:** a release-pipeline check fails a release that has no `fixtures/<version>/`.-Commit hashes are filled in when the phase ends.+Phase 1 ended on 2026-10-24. Commit 49177427 came from another session.| Step | Commit | Notes ||---|---|---|-| 1.1 Pinned CLI | — | CLI v1.9.1 (server 1.32.0) is pinned in `server/config/temporal_cli.json` and verified with pooch's `known_hash`. It is installed under `packages/temporal/<version>/`, and every other CLI file there is deleted. `_install.prepare_database` deletes a `temporal.db` without the pinned server minor's marker. The dev server also gets fairness and one task-queue partition from the pin's `dynamic_config` (T17); the marker records the partitions, and a database created with others starts from new. Commit 49177427 (another session) had replaced the deletion with keep-and-refuse; that was reverted after T5. **Release note:** the first release with the pin starts every install on an empty Temporal database, so every employee needs Reset and Start. |-| 1.2 CI | — | `backend-tests` runs on ubuntu, windows and macos with full history. The app's installer puts the pinned CLI into a throwaway `DATA_DIR` and exports it as `TEMPORAL_TEST_CLI`, so the real-server tests run instead of skipping. Locally on Windows, all 8 of those items passed against the pinned binary. The first CI run on macOS is the first on that OS. A gate step runs the replay and identity tests before the full suite, with `TEMPORAL_REQUIRE_NATIVE_REPLAY=1`. |-| 1.3 Fixture recorder | — | `server/tests/temporal/record_fixtures.py` records all 14 production types along 21 scenarios on a throwaway dev server, from a git archive of `--source-commit`. The 0.3.1 set, first recorded on CLI 1.8.0, was recorded again on the pinned 1.9.1 from the same commit: 51 histories. |-| 1.4 Replay gate | — | `test_replay_corpus.py` replays every set with `workflow_registry.replay_workflows()`, checking source provenance and coverage of all 14 types. A release-pipeline check fails a version without `fixtures/<version>/` or with a set recorded on another CLI than its pin. |-| 1.5 Golden identity test | — | `fixtures/workflow_identities.json` holds the release's commands and ids, handler names and plugin activity names. The test fails on unreviewed command changes, a patched marker that vanishes without a release of `deprecate_patch`, and a renamed or removed handler. |-| 1.6 temporalio 1.34.0 | — | The floor is `temporalio>=1.34.0` (RFC-0004 D5), and only temporalio moved in `server/uv.lock`. No code had to change:<br>• nothing used `PayloadSizeWarning`, removed in 1.31 when payload limits moved into the SDK's core;<br>• every `workflow.continue_as_new` call is in a `@workflow.run` method. From 1.33, continuing as new from an Update handler fails the workflow task.<br>On 1.34.0 the 0.3.1 corpus replays, and the full server suite shows no failure caused by the upgrade. From the next release, new patch markers ship dormant for one release (RFC-0006 §18.9). |+| 1.1 Pinned CLI | b357b22e, 7caa49e4, 1ffb3f69 | CLI v1.9.1 (server 1.32.0) is pinned in `server/config/temporal_cli.json` and verified with pooch's `known_hash`. It is installed under `packages/temporal/<version>/`, and every other CLI file there is deleted. `_install.prepare_database` deletes a `temporal.db` without the pinned server minor's marker. The dev server also gets fairness and one task-queue partition from the pin's `dynamic_config` (T17); the marker records the partitions, and a database created with others starts from new. Commit 49177427 (another session) had replaced the deletion with keep-and-refuse; that was reverted after T5. **Release note:** the first release with the pin starts every install on an empty Temporal database, so every employee needs Reset and Start. |+| 1.2 CI | a9ee6675, 49177427, 449bfb6c | `backend-tests` runs on ubuntu, windows and macos with full history. The app's installer puts the pinned CLI into a throwaway `DATA_DIR` and exports it as `TEMPORAL_TEST_CLI`, so the real-server tests run instead of skipping. Locally on Windows, all 8 of those items passed against the pinned binary. The first CI run on macOS is the first on that OS. A gate step runs the replay and identity tests before the full suite, with `TEMPORAL_REQUIRE_NATIVE_REPLAY=1`. |+| 1.3 Fixture recorder | 46994a80, 49177427, c182f6f9 | `server/tests/temporal/record_fixtures.py` records all 14 production types along 21 scenarios on a throwaway dev server, from a git archive of `--source-commit`. The 0.3.1 set, first recorded on CLI 1.8.0, was recorded again on the pinned 1.9.1 from the same commit: 51 histories. |+| 1.4 Replay gate | 49177427, 72d460a9 | `test_replay_corpus.py` replays every set with `workflow_registry.replay_workflows()`, checking source provenance and coverage of all 14 types. A release-pipeline check fails a version without `fixtures/<version>/` or with a set recorded on another CLI than its pin. |+| 1.5 Golden identity test | 49177427, d3f8e873 | `fixtures/workflow_identities.json` holds the release's commands and ids, handler names and plugin activity names. The test fails on unreviewed command changes, a patched marker that vanishes without a release of `deprecate_patch`, and a renamed or removed handler. |+| 1.6 temporalio 1.34.0 | 2ea28bdf | The floor is `temporalio>=1.34.0` (RFC-0004 D5), and only temporalio moved in `server/uv.lock`. No code had to change:<br>• nothing used `PayloadSizeWarning`, removed in 1.31 when payload limits moved into the SDK's core;<br>• every `workflow.continue_as_new` call is in a `@workflow.run` method. From 1.33, continuing as new from an Update handler fails the workflow task.<br>On 1.34.0 the 0.3.1 corpus replays, and the full server suite shows no failure caused by the upgrade. From the next release, new patch markers ship dormant for one release (RFC-0006 §18.9). |## 25. Defaults chosen and open items