Repository navigation
feat(t27os): NVIDIA profile for OpenCode panes in the key master (Refs #12983) - #12986
Merged
Merged
Conversation
…outside the chain (Refs #12983) PR_NVIDIA sits above the four chain profiles; agent_profile_ok_stored admits it for OpenCode only while an NVIDIA key is stored. The inline config names provider nvidia alone (enabled_providers), holds the key by reference and no model id. Tests pin key present/absent, paid provider, unknown profile and the key-free config text. Refs #12975, #12977. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Contributor
This was referenced Oct 31, 2026
Collaborator
Author
|
CI triage at head eb81675: the two red checks are inherited from master, not from this change (it touches specs/t27os/keys.t27 only).
Nothing pushed for these two. 🤖 Generated with Claude Code |
Collaborator
Author
|
Triage of the third CI wave on this PR (head eb81675). This PR changes only specs/t27os/keys.t27. Caused by this PR (1):
Not caused by this PR (2):
No fix was pushed for these two. Refs #12983 🤖 Generated with Claude Code |
This was referenced Oct 31, 2026
Merged
Contributor
PR DashboardGenerated at: 2026-10-31 08:56:35 UTC
Summary
Seal Status
|
Collaborator
Author
|
Triage of the fourth CI wave on this PR (head a4061dc = master 5550045 merged into the branch). This PR changes only specs/t27os/keys.t27. No fix was pushed. Caused by this PR (1 seal, shown by 2 checks):
Not caused by this PR (the same checks fail on master tip 5550045):
Refs #12983 🤖 Generated with Claude Code |
…12983) One seal file. Saved with the post-#12824 t27c at ~/t27-lanes/t27c-bc19b54e6 (base bc19b54), on the owner's word for exactly one save, run from a shell cell. The spec hash and the four generated-code hashes now describe keys.t27 at this head; tests 115 -> 117. Refs #12983 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Contributor
PR DashboardGenerated at: 2026-10-31 09:17:54 UTC
Summary
Seal Status
|
Collaborator
Author
|
Triage of the sixth CI wave on this PR (head 25ff6dd, the T27osKeys seal re-save). No fix was pushed. Caused by this PR: nothing left. Still red, all the same on master tip (not caused by this PR, nothing blessed):
The PR stays a draft. Nothing here needs a change in specs/t27os/keys.t27. Refs #12983 🤖 Generated with Claude Code |
This was referenced Oct 31, 2026
Contributor
PR DashboardGenerated at: 2026-10-31 13:22:33 UTC
Summary
Seal Status
|
This was referenced Oct 31, 2026
This was referenced Nov 9, 2026
Co-authored-by: Dmitrii Vasilev <gHashTag@users.noreply.github.com>
This was referenced Nov 9, 2026
Contributor
PR DashboardGenerated at: 2026-11-09 22:06:28 UTC
Summary
Seal Status
|
This was referenced Nov 9, 2026
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.
Refs #12983 (part of #12975; the lane law is draft PR #12977). Not a closing PR: the runtime success criterion (a pane answering one probe) is not done here.
What changed (specs/t27os/keys.t27 only)
PR_NVIDIA = 4, deliberately ABOVE the four chain profiles. Every table indexed byPROFILE_COUNT(names, tiers, models, base URLs, chain walk, Claude pane) is untouched; a test pinsPR_NVIDIA == PROFILE_COUNT, so growing the tables is a decision, not a collision.agent_profile_ok(a, pr):PR_NVIDIAis OpenCode-only (kind check, as before no key in sight).agent_profile_ok_stored(a, pr, stored): the NVIDIA profile starts only while an NVIDIA key row exists (bitP_NVIDIA). A key of any other provider, a paid one included, never stands in. Unknown profile or agent: no.opencode_config(PR_NVIDIA, 0, out): one inline config naming providernvidiaalone (enabled_providers:["nvidia"]), the base URL, and the key by reference ({env:NVIDIA_API_KEY}). NO model id and no model table: the host takes ids from the live listing; a listed model is not an entitlement (access-map 2026-10-26: listing 200, chat 403). Step != 0 returns 0.profile_provider(pr)(new) backsagent_var_name, soPROFILE_PROVIDER[pr]is never indexed out of range;oai_base_urlanswersPR_NVIDIA.Deviation from the issue text (FR-001)
agent_profile_ok(a, pr)has no stored-mask parameter, and adding one would force edits inpty_logic.t27, outside the boundary. So the kind check stays inagent_profile_okand the keyed gate isagent_profile_ok_stored.Receipts
t27c check specs/t27os/keys.t27-> Typecheck OK (0 errors, 1 warning, the same one as on master). Added lines are ASCII; no key value ornvapistring in the spec; lefthook foreign-budget, own-language and gitleaks passed on commit and push.test-reportand the tests themselves (no heavy runs on the Mac; CI is the lab). No probe was run and no key was used.specs/t27os/opencode_free.t27in draft PR feat(t27os): OpenCode free pool as a second executor, law in .t27 (Refs #12975) #12977.Follow-ups outside the boundary
pty_logic.t27: the< PROFILE_COUNTguards (about lines 741 and 1060) and the spawn path must useagent_profile_ok_storedandPR_NVIDIA;pty_server.t27wiring.AGENT_STRIPSshould addNVIDIA_API_KEY;AGENT_PROVIDERS[3]needs the OpenCode bit for nvidia.opencode_free.t27deciding admission.🤖 Generated with Claude Code
Note
Low Risk
Spec-only decisions in
keys.t27with tests; no runtime spawn or key handling changed in this PR.Overview
Adds
PR_NVIDIAas a fifth pane profile inspecs/t27os/keys.t27, wired into the existingPROFILE_COUNTtables (name, tier, provider bitP_NVIDIA, OpenAI base URL) so OpenCode—not Claude/Crush—can target integrate.api.nvidia.com.Admission splits into
agent_profile_ok(agent/kind only) and newagent_profile_ok_stored, which allows the NVIDIA profile only when an NVIDIA key row is present; other providers’ keys do not substitute.opencode_configforPR_NVIDIAemits a minimal inlineOPENCODE_CONFIG_CONTENT: singlenvidiaprovider, base URL, and{env:NVIDIA_API_KEY}—no baked-in model id (host picks from live listings). Newprofile_providerkeepsPROFILE_PROVIDERindexing safe for helpers likeagent_var_nameandoai_base_url.Includes tests for stored-key gating and config shape. Runtime spawn/probe wiring (
pty_logic.t27, strips) is explicitly out of scope here.Reviewed by Cursor Bugbot for commit 288e059. Bugbot is set up for automated code reviews on this repo. Configure here.