Simplify harness trigger and async run::start#174
Draft
ytallo wants to merge 10 commits into
Draft
Conversation
Console calls harness::trigger with a flat payload; harness always starts run::start. Remove run::start_and_wait, cwd state keys, and initial agent::events replay at kickoff so the FSM owns turn lifecycle.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Important Review skippedDraft detected. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
skill-check — worker0 verified, 12 skipped (no docs/).
Three for three. Nicely done. |
Replace manual typeof/Record parsing with shared TurnStateWriteEventSchema and parseTurnStateWrite, matching the run-start/trigger refactor style.
Each state-trigger adapter now exports register() with inline function ids, matching run-start.ts. register.ts only composes modules; exported ID constants removed.
… the register condition.
…ng is type-enforced. Share Zod turn_state write parsing between adapters; add handler tests for direct invokes that bypass the engine condition gate.
…ype safety - Simplified `parseStepableWrite` to only return `session_id`, removing the state from its output. - Inlined the handling of stepable writes directly in `handleStepableRecordWrite`, eliminating the fallback to durable publish. - Updated tests to reflect changes in the `parseStepableWrite` output and removed obsolete tests related to fallback logic. - Introduced `StepPayloadSchema` for consistent payload validation in the `turn::step` function, ensuring only valid session IDs are processed. - Added new tests for `StepPayloadSchema` to validate input shapes and error handling. These changes enhance clarity and maintainability of the turn orchestrator's state management logic.
Replace turn::step_requested publishes with queue enqueues in approval resume and abort handling, add turn::should_step filtering for unknown or terminal sessions, and tighten get_state with Zod payload parsing.
Replace the monolithic turn::step subscriber with turn::{state} functions
enqueued on turn-step, consolidate UI turn_state_changed emits into
persistence saves, and merge assistant/function states for a simpler FSM.
Reconcile the turn-orchestrator per-state queue refactor with main's harness-node/ -> harness/ rename and parallel changes: - Map the refactor (per-state handlers, turn-step FIFO queue, Zod boundary parsing, simplified harness::trigger) onto the harness/ tree. - Adopt main's idempotency guards: function_result dedup lands in function-execute.ts via the rename merge; the assistant-message guard is ported into the split assistant-finished.ts handler. - Take main's new providers (kimi/lmstudio/llamacpp), functionNotFoundHint hint, and system-prompt guidance (auto-merged, untouched by the refactor). - Drop on-record-written/on-terminal/on-turn-state-changed/subscriber/ transitions removed by the refactor; keep main's provider tests. - Update functions error-result test for the new executedCalls clear, and re-add idempotency regression tests adapted to the split handlers. typecheck clean; 826 tests pass.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
harness::triggerwith a flat payload only; harness-node always invokesrun::start(no client-suppliedfunction_id).run::startuses Zod defaults, drops unusedcwd/cwd_hashfrom the run payload and iii state, and no longer replays syntheticagent::eventsat kickoff.run::start_and_waitpath (on-terminalwaiter, policy entry,sync_default_timeout_msconfig).Test plan
pnpm run typecheckinharness-nodepnpm testinharness-node(494 tests)Notes
docs/flow HTML was left out of this PR; say if you want it included.run::start_and_waitonmain; out of scope here.