Conversation
Parse scheduled task wake reasons, validate current activations, and surface live trigger candidates so the slow brain can start the right task.
The live runtime was validating due and trigger activations against the SDK default project, so staging sessions saw empty machine-state rows even when Orchestra had projected them correctly. Pin machine-state reads to the Unity project and log wake-reason replay and due-task rejection outcomes so live failures are visible.
Add the headless offline runner, enforce offline task contracts, and rearm recurring scheduled tasks from the previous occurrence. This also mirrors due and trigger context into the voice fast brain before the opening greeting.
Move task due, trigger matching, and fast-brain wake context into a dedicated domain module so event handlers stay focused on event routing. Keep the existing communication handling intact while making the task activation path easier to read.
Restore the pre-refactor room-name behavior when WhatsApp permission is granted before an assistant id is populated. Add a focused regression test so the handler stays behavior-preserving.
Use the richer wake contract to give the slow brain and fast brain clearer scheduled and triggered task context. This keeps wake handling more colleague-like without exposing offline work or internal task-machine jargon.
The startup greeting should stay social even when buffered task wake notifications are already present in history. Move the greeting-specific guardrail into the prompt builders and align the first-turn eval with the real fast_brain_greeting path so the test matches production.
Task activations and runs now live in Assistants, so Unity should query the same project for due validation and run updates.
Add a public deploy-runtime seam with local defaults so Unity can stay local-first while hosted session assignment, metrics, and job lifecycle logic move behind private runtime backends.
Capture scheduled and triggered provenance before live execution begins so Unity can create or adopt the durable run row at task start and patch terminal state back to Orchestra. Align queue and offline execution updates with the assistant-scoped run contract and cover the new behavior in task scheduler and activation tests.
Point the ConversationManager fast-brain defaults and prompt test fixture at gpt-5.4-mini so voice calls use the newly supported low-latency model by default.
Clarify Unity's positioning as persistent AI colleagues and move the demo, architecture, and evaluator entry points above the fold so the repo is easier to assess quickly.
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
2 participants
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.
Merge all staging changes to main.