workstream: default discover model to sonnet - #316
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
pigeon workstream discoverdefaults to haiku. On workspaces with many overlapping workstreams, haiku tends to over-decompose — producing many fine-grained workstreams that share vocabulary and confuse downstream routing.Solution
Default to sonnet for both
workstream discoverandworkstream replay. Sonnet auto-consolidates duplicates during discovery, producing fewer but better-separated workstreams. Per-call cost rises by a few cents; discovery is run rarely (workstream creation/refresh, not per-message), so total impact is negligible.Context
Validated empirically on three workspaces of varying size and structure. Sonnet ties haiku on cross-workspace mean recall under the focus-filter routing harness, but its failure mode is gentler: where haiku over-decomposes (and the embedder can't tell the resulting workstreams apart), sonnet consolidates correctly. On clean small workspaces sonnet is a small regression; on messy larger ones it's a clear win. Opus is consistently worse on recall and produces over-conservative routing — not recommended.
Full methodology and three-way comparison are in PR #311.