fix(opencode): cap 70B NIM output and lock released notes by cursor[bot] · Pull Request #545 · ContextualWisdomLab/scopeweave

cursor · 2026-08-16T16:53:32Z

Summary

Land the NVIDIA NIM-only local OpenCode development config on current develop, then apply the two remaining current-head findings from #543.

#543 def2dfb already restores the published 1.0.0 / 1.0.1 changelog notes and switches opencode.jsonc to NVIDIA NIM with {env:NVIDIA_API_KEY}. This successor keeps that slice and adds:

meta/llama-3.3-70b-instruct limit.output set to 4096, matching NVIDIA NIM max_tokens 1–4096 (NIM 70B infer reference)
unit assertion for that provider cap
tests/unit/changelog-release-notes.test.mjs so a later commit cannot delete published 1.0.0 / 1.0.1 notes again
changelog article fix: "an NVIDIA NIM-only"

The default 49B path stays at documented max_tokens 1–65536. Organization OpenCode / Strix / merge-scheduler workflows stay in ContextualWisdomLab/.github. Local binding remains {env:NVIDIA_API_KEY}; org CI continues to map NVIDIA_NIM_API_KEY into that process binding. Do not restore github-models, STRIX_GITHUB_MODELS_TOKEN, or COPILOT_GITHUB_TOKEN.

Prefer this PR over #543 and over the unopened c9cd78c branch. After merge, continue the calendar grant stack (#541 / #539) and Clearfolio production adapter (#489) rather than opening another NIM-config PR.

Test plan

node tests/unit/opencode-config.test.mjs
node tests/unit/changelog-release-notes.test.mjs
npm run test:unit on CI exact head

Keep the published 1.0.0/1.0.1 changelog notes locked and align the optional 70B candidate with NVIDIA NIM's documented 1-4096 max_tokens range so OpenCode cannot request an over-limit completion. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

seonghobae

Reviewed the exact current head. The repository-local provider boundary is limited to NVIDIA NIM, the configured 49B/8B/70B identifiers and token limits align with NVIDIA's current NIM API references, GitHub Models/Copilot token bindings are removed, organization workflow ownership remains external, and focused regressions lock both the provider contract and released changelog notes. Approval is exact-head only; Draft state and all queued CI/security/organization gates must still complete before integration.

cursor · 2026-08-16T19:08:39Z

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

seonghobae · 2026-08-16T21:07:11Z

@coderabbitai review

coderabbitai · 2026-08-16T21:07:17Z

⚠️ Action not completed

Review rate limited.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

coderabbitai · 2026-08-16T21:07:20Z

Warning

Review limit reached

@seonghobae, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 38 minutes

Limit details: You’ve used all 1 included review currently available under your plan.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: babe6145-4f2e-42b7-ad7b-5cea5ec8f175

📥 Commits

Reviewing files that changed from the base of the PR and between 44e7903 and ef9f9a9.

📒 Files selected for processing (6)

AGENTS.md
CHANGELOG.md
opencode.jsonc
package.json
tests/unit/changelog-release-notes.test.mjs
tests/unit/opencode-config.test.mjs

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

seonghobae · 2026-08-16T23:06:43Z

@coderabbitai review

coderabbitai · 2026-08-16T23:06:48Z

⚠️ Action not completed

Review rate limited.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

seonghobae

Reviewed the exact current head against protected develop and current NVIDIA primary API references. The bounded diff is ahead-only on develop, keeps organization review workflow ownership outside the repository, removes GitHub Models/Copilot credential bindings from the local OpenCode config, and matches NVIDIA's hosted limits for the configured 49B, 8B, and 70B candidates (including 49B max_tokens 65536 and 70B/8B max_tokens 4096). The focused regressions lock the provider contract and preserve published changelog notes. This approval is source-review evidence for exact head ef9f9a9 only; the current Strix annotation reports backend unavailability, so it remains non-passing security evidence and must be rerun authoritatively before merge.

seonghobae · 2026-08-17T05:43:31Z

@opencode-agent review

Please review exact current head ef9f9a9912b37946ee340aef26f7f2d62a03e32d only against live protected develop@44e7903cf8891c65410f7fc6ca5144de3fdb5185. Submit a formal exact-head APPROVED or CHANGES_REQUESTED review; do not treat the existing opencode-review check/status as a verdict and do not transfer predecessor-head evidence. Scope is the six-file repository-local NVIDIA NIM development-config/changelog regression slice; organization review workflow ownership remains in ContextualWisdomLab/.github.

Keep the local OpenCode binding as NVIDIA_API_KEY and assert that AGENTS.md still documents the organization secret injection split. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

seonghobae

Exact-head review of 2bdd2159664636a0d9c5c5b64c230ae2e6696716 against live develop@44e7903.

The NIM-only local OpenCode slice is unchanged and still correct: enabled_providers is nvidia-nim only, the client binding stays {env:NVIDIA_API_KEY}, the org secret remains NVIDIA_NIM_API_KEY, 70B output stays at 4096, and GitHub Models / Copilot tokens are absent. The new commit only locks that org-secret mapping in tests/unit/opencode-config.test.mjs.

Local verification on this head:

node tests/unit/opencode-config.test.mjs PASS
node tests/unit/changelog-release-notes.test.mjs PASS

Last pusher on this head is Cursor Agent, not the reviewer, so this approval can satisfy require_last_push_approval.

Analyze (python) failed during CodeQL init with GitHub HTTP 503, not a source finding. Retrigger exact-head default-setup checks after API and Actions recovered. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

seonghobae

Exact-head review of c8eb58a86410c4f0d6e3be575a13d82d77826dba against live develop@44e7903.

This commit is empty and only retriggers required CodeQL after Analyze (python) failed in CodeQL init with GitHub HTTP 503 during the 2026-08-17 GitHub.com outage. The NIM-only local OpenCode slice is unchanged from 2bdd215:

enabled_providers is nvidia-nim only
client binding stays {env:NVIDIA_API_KEY}
org secret remains NVIDIA_NIM_API_KEY
70B output stays at 4096
GitHub Models / Copilot tokens are absent

Last pusher on this head is Cursor Agent, not the reviewer, so this approval can satisfy require_last_push_approval. Predecessor 2bdd215 required org workflows (Strix, OpenCode, Noema, Semgrep, Security Scan) were green; this approval does not transfer those checks. Merge still requires exact-head Analyze (python) and the other required gates on c8eb58a.

Reconcile PR #468 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the editor-validation and fullwidth CSV hardening slice.

Reconcile PR #490 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the rerender-safe focus restoration and exact coverage evidence.

Reconcile PR #523 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the exact-head Server Tests, CodeQL, OSV live-base evidence, and workflow regression contract.

Reconcile PR #493 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the bounded Clearfolio production configuration and provider-origin hardening slice.

Reconcile PR #495 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the metadata-only render path, immutable badge shells, and 5,000-row browser benchmark.

Reconcile PR #496 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving server-derived tenant cost attribution and contextual-orchestrator auto mode.

…nfig Reconcile PR #497 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the audited, explicit-ID orphan workflow cleanup controls.

Reconcile PR #500 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the fail-closed schema migration ledger and catalog-first legacy column upgrades.

Reconcile PR #503 onto current protected develop, preserving the canonical runtime authority rewrite while retaining the protected NVIDIA NIM/OpenCode development boundary introduced by #545.

Reconcile PR #508 onto current protected develop, retaining #545 OpenCode/NIM configuration and release-note regressions while preserving the computeTaskMetrics hot-path optimization and immutable-base benchmark evidence.

Reconcile PR #509 onto current protected develop, retaining #545 NVIDIA NIM/OpenCode configuration and published-release-note regressions while preserving the bounded metric-card accessibility changes.

Reconcile the hierarchy-domain root onto current protected develop, preserving #545 NVIDIA NIM/OpenCode configuration and released-note regressions while retaining the work-item hierarchy unit/coverage registrations.

seonghobae and others added 10 commits August 17, 2026 01:40

test(opencode): require NVIDIA NIM-only local config

b0ea760

test(opencode): execute NIM config regression in unit suite

4b2421e

test(opencode): require live hosted NIM fallback set

6f8fbf5

fix(opencode): use NVIDIA NIM-only local development config

15f80bf

docs(opencode): keep local NIM config behind central workflow ownership

2bb47e2

docs(changelog): record NIM-only OpenCode development config

aa4a018

test(opencode): use hosted 8B NIM model for small tasks

559a77d

fix(opencode): use hosted 8B NIM model for small tasks

f3bf4bb

fix(changelog): preserve existing release history

def2dfb

cursor Bot requested a review from seonghobae August 16, 2026 16:53

cursor Bot mentioned this pull request Aug 16, 2026

fix(opencode): move local development config to NVIDIA NIM #543

Closed

seonghobae previously approved these changes Aug 16, 2026

View reviewed changes

seonghobae marked this pull request as ready for review August 16, 2026 19:08

merge(develop): preserve toast accessibility with NIM config

ef9f9a9

seonghobae dismissed their stale review via ef9f9a9 August 16, 2026 21:06

seonghobae self-requested a review August 16, 2026 21:55

seonghobae previously approved these changes Aug 17, 2026

View reviewed changes

seonghobae mentioned this pull request Aug 17, 2026

feat(automation): run Scopeweave hourly NVIDIA NIM review repair ContextualWisdomLab/.github#1078

Closed

test(opencode): lock org NVIDIA_NIM_API_KEY mapping in AGENTS.md

2bdd215

Keep the local OpenCode binding as NVIDIA_API_KEY and assert that AGENTS.md still documents the organization secret injection split. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

cursor Bot dismissed seonghobae’s stale review via 2bdd215 August 17, 2026 17:23

seonghobae approved these changes Aug 17, 2026

View reviewed changes

seonghobae merged commit 1fadec0 into develop Aug 17, 2026
37 checks passed

seonghobae deleted the cursor/bc-ac8b4bcb-2f2c-47ff-b86b-b4d0f2517371-ed01 branch August 17, 2026 18:56

seonghobae mentioned this pull request Aug 17, 2026

fix(billing): bound Stripe Checkout provider trust #507

Open

seonghobae mentioned this pull request Aug 17, 2026

fix(ci): execute Server Tests on exact PR heads #523

Open

Navigation Menu

Uh oh!

fix(opencode): cap 70B NIM output and lock released notes - #545

fix(opencode): cap 70B NIM output and lock released notes#545
seonghobae merged 13 commits into
developfrom
cursor/bc-ac8b4bcb-2f2c-47ff-b86b-b4d0f2517371-ed01

cursor Bot commented Aug 16, 2026

Uh oh!

seonghobae left a comment

Uh oh!

cursor Bot commented Aug 16, 2026

Uh oh!

seonghobae commented Aug 16, 2026

Uh oh!

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading

Review limit reached

Uh oh!

seonghobae commented Aug 16, 2026

Uh oh!

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading

Uh oh!

seonghobae left a comment

Uh oh!

seonghobae commented Aug 17, 2026

Uh oh!

seonghobae left a comment

Uh oh!

seonghobae left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

cursor Bot commented Aug 16, 2026

Summary

Test plan

Uh oh!

seonghobae left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot commented Aug 16, 2026

Uh oh!

seonghobae commented Aug 16, 2026

Uh oh!

coderabbitai Bot commented Aug 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot commented Aug 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review limit reached

Uh oh!

seonghobae commented Aug 16, 2026

Uh oh!

coderabbitai Bot commented Aug 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

seonghobae left a comment

Choose a reason for hiding this comment

Uh oh!

seonghobae commented Aug 17, 2026

Uh oh!

seonghobae left a comment

Choose a reason for hiding this comment

Uh oh!

seonghobae left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading