Skip to content

Validate Grok 4.6 live behavior and close stability evidence - #4

Open
hffmnnj wants to merge 8 commits into
fix/cursor-request-livenessfrom
test/cursor-live-validation
Open

Validate Grok 4.6 live behavior and close stability evidence#4
hffmnnj wants to merge 8 commits into
fix/cursor-request-livenessfrom
test/cursor-live-validation

Conversation

@hffmnnj

@hffmnnj hffmnnj commented Aug 12, 2026

Copy link
Copy Markdown
Owner

Summary

  • validate qualified Grok 4.6 routing in an isolated authenticated OpenCode process
  • confirm Cursor emits turn completion while HTTP/2 remains writable
  • add opt-in sanitized diagnostics for wire model and terminal transport state
  • run the fully stacked regression gates and close wiring evidence
  • document model variants, pricing, timeouts, diagnostics, and the final root cause

Verification

  • npx tsc -p tsconfig.json --noEmit
  • bun test/smoke.ts
  • bun run build
  • git -c core.pager=cat diff 585b7e1..HEAD --check

This change is stacked on fork PR #3.

Document the safe tmux validation attempt and its authentication-limited outcome without retaining raw diagnostics or secrets.
Record isolated model routing and lifecycle observations while preserving the remaining A1 and A7 limitations.
Add opt-in, allowlisted diagnostics for the Cursor wire model and turn-ended bridge state. This makes live validation observable while keeping payloads, prompts, headers, tokens, and transport addresses out of logs.\n\nChanges:\n- Emit outbound wire model identifiers only under diagnostic mode\n- Capture parent-observed bridge state before turn-ended cleanup\n- Assert diagnostic schemas and redaction in smoke coverage
Record the authenticated outbound model id and writable transport state observed at turnEnded.
Capture the fully stacked typecheck, smoke, and build results plus script inspection and model no-regression evidence.
Remove trailing whitespace from the final gate record.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant