Improve first-run generation defaults and limit request concurrency - #349
Merged
Merged
Conversation
zurawiki
added this pull request to stack #350
September 24, 2026 14:33
zurawiki
marked this pull request as ready for review
September 24, 2026 14:37
zurawiki
force-pushed
the
qol/better-defaults
branch
from
September 24, 2026 14:37
d43a4e1 to
2cbd58f
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Make first-run generation quieter and bounded. Skip empty and fully ignored changes before constructing the API client, leaving the commit message unchanged even without credentials. Limit file-summary requests to four at once, order summaries by filename, and propagate per-file failures instead of silently substituting empty summaries. Add bun.lock and uv.lock to default exclusions. Routine progress and exclusions are available through --verbose.
Update the default model to gpt-6-luna with low reasoning effort to prioritize speed and a 2,048-token combined reasoning/output limit per request; retain explicitly configured models and avoid sending those model-specific options to other models. Shorten default prompts to a <=60-character title (leaving room for the conventional prefix) and one to three factual bullets, without invented rationale or test claims.
Stacked on #348. Moves filtering out of per-file tasks so it can run before client setup, removes unbounded task spawning and silent summary-error handling, and replaces the promotional generation banner with diagnostic logging.
Model rationale: GPT-6 Luna is documented for focused, high-volume tasks and supports low reasoning effort. Published standard short-context pricing is $0.10 input / $0.50 output per million tokens, versus $0.10 / $0.40 for the previous GPT-4.1 nano default. These are documentation-based choices; no live quality, latency, or account-availability comparison has been performed. Keep this PR draft until that evaluation is satisfactory.
Validation: focused tests cover concurrency, aborting incomplete summaries, empty input without requests, default-model payload options, and unchanged payloads for custom models. CLI regressions cover no-key empty/all-ignored changes and unchanged message files.
just lint build testpassed, including the CLI end-to-end suite and all 15 Rust tests.Follow-up validation: request-payload regressions and
just lintpassed after selecting low reasoning. The existing test now enforces that default and preserves custom-model payloads. Live latency remains unmeasured.