Repository navigation
feat(opencode): reduce Sol reviewer reasoning to high - #133
Merged
Merged
Conversation
Chart CI passedRepository hygiene and Helm validation passed. View the workflow run. |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Change only
adversarial-code-reviewerandinfra-security-reviewerfrom GPT-6.1 Solxhightohigh. Preserve their models, prompt bodies and permissions; retain all ten primary agents, global Sol fallback, other workers, and GLM-Flashmax. Bump immutable chartopencode-serverfrom0.5.7to0.5.8; appVersion2.0.22and API selector2stay unchanged.Owner-approved risk/cost starting policy following the reasoning-effort investigation, not measured optimality, equivalent review quality, or demonstrated savings. OpenAI recommends reserving xhigh for evaluated benefits; GLM effort labels and intended maximum-reasoning role are distinct.
Fixes: N/A — owner-requested reasoning policy change.
Type of change
Validation
test,detect, andpackage opencode-serversucceeded at the reviewed HEAD. The post-publish GitOps updater was correctly skipped on this PR. CI is the validation environment; no local checks run.Expected checks:
test,detect,package opencode-server(repository hygiene, strict Helm lint/render, model/variant source/render assertions, baseline parity and archive packaging). Add a separate exact two-header transform after historical reasoning and Terra restoration; preserve both historical baseline SHAs and transforms. Fixtures cover allowlisted changes, unchanged bodies/unrelated files, and missing/wrong/duplicate selections.Pre-PR reviews at
a9d5617b2eb6b8c8b07707e7b7cbae9fdb226100: adversarial ADVANCE, infrastructure-security ADVANCE, DevOps ADVANCE, QA COVERED. Release assessment found no version/packaging defect in supplied evidence; all independent review gates and PR checks are now complete. Merge/publication still require explicit owner confirmation. These are independent agent assessments, not GitHub approvals. No Critical/High finding remains.Impact and rollout
Canonical producer:
makeitworkcloud/charts, targetmain. Eight paths: Makefile; chart metadata; two reviewer agent definitions; memory-pilot tests; README; architecture and memory-pilot guides. No image, provider, model, workflow, permission, secret, resource or topology change.0.5.8to GHCR.makeitworkcloud/kustomize-clusterpin PR with auto-merge enabled and required checks. At verified consumer revision2edd55319072c9bc52ecece692c36925c28bd715, Application pin and exact chart/API assertion are0.5.7/2. The future consumer PR must update pin and exact assertion atomically to0.5.8/2; pin-only automation cannot pass that check. No consumer edit is included.0.5.7/API2; not an image downgrade.Safety and secrets
AI-assisted implementation and independent review. Merge, publication, consumer adoption, and live rollout require separate confirmation.