refactor: tighten architect/plan-reviewer contract for lower-tier orchestration#67
Draft
IgorWnek wants to merge 11 commits into
Draft
Conversation
Contributor
Author
|
FYI: plan.example.md work is not finished here, it'll be subject to further changes in this PR |
…h-creating-implementation-plans skill
…viewer and engineering manager
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What & why
The implementation orchestration loop now runs on a lower-tier orchestrator model that delegates high-leverage planning to two top-tier advisors:
tsh-architect(plan author) andtsh-plan-reviewer(plan challenger). The contract between those two agents had gaps that made the review→revise loop hard for the orchestrator to route reliably:tsh-architectwas still unstructured prose and mixed WHO (behavior) with HOW (workflow mechanics), and had no explicit contract for consuming review feedback.tsh-plan-reviewercarried an impossible self-routing instruction ("always send the report to the architect") even though the orchestrator — not the reviewer — owns routing. Its verdict was also not machine-readable, so the orchestrator had to infer next steps from prose.This PR restructures and hardens that contract without changing what the loop does. Severity levels, verdicts (
APPROVED/REVISIONS NEEDED), the max-3-iteration cap, the "skip re-review if approved and unchanged" rule, and.plan-review.mdas source of truth are all preserved.