Skip to content

Add v3 workflow gauntlet proof protocol#18

Merged
graphanov merged 1 commit into
mainfrom
proof/workflow-gauntlet-v1
Jun 3, 2026
Merged

Add v3 workflow gauntlet proof protocol#18
graphanov merged 1 commit into
mainfrom
proof/workflow-gauntlet-v1

Conversation

@graphanov

Copy link
Copy Markdown
Owner

Summary:

  • add a neutral v3 workflow gauntlet schema, frozen-candidate fixture, validator, and docs
  • require unstructured, minimal-checklist, and structured-workflow lane roles
  • keep mechanical, visual, workflow, trajectory, token-cost, and evidence tracks separate

Guardrails:

  • no Open Scaffold-specific contract
  • no Lane C
  • no evidence-volume scoring
  • token-cost stays separate from mechanical scoring
  • visual ranking requires valid native capture

Verification:

  • git diff --check
  • python3 scripts/validate_v3_workflow_gauntlet.py
  • python3 scripts/validate_v3_schemas.py
  • python3 scripts/validate_v3_workflow_scenarios.py
  • python3 scripts/validate_v3_campaign_protocol.py
  • python3 scripts/scan_changed_public_safety.py
  • python3 scripts/run_v3_token_telemetry_smoke.py
  • python3 scripts/run_v3_visual_smokes.py --visual-out v3/fixtures/visual/out
  • python3 scripts/run_v3_mechanical_smokes.py
  • python3 scripts/run_v3_feedback_parity_smoke.py
  • python3 scripts/validate_v3_visual_packages.py v3/fixtures/visual/out/visual-package.json
  • cargo fmt --all --check
  • cargo build --workspace --quiet
  • cargo clippy --workspace -- -D warnings
  • cargo test --workspace --quiet

Constraint: Private Gate 6 showed mechanical ties, visual blockers, and token waste; benchmark changes must stay workflow-agnostic.

Rejected: Re-run Open Scaffold lane immediately | the existing protocol cannot distinguish scaffold value from cheap checklist discipline.

Confidence: high

Scope-risk: moderate

Directive: Do not convert this fixture into an Open Scaffold-specific scoring contract.

Tested: git diff --check; python3 scripts/validate_v3_workflow_gauntlet.py; python3 scripts/validate_v3_schemas.py; python3 scripts/validate_v3_workflow_scenarios.py; python3 scripts/validate_v3_campaign_protocol.py; python3 scripts/scan_changed_public_safety.py; python3 scripts/run_v3_token_telemetry_smoke.py; python3 scripts/run_v3_visual_smokes.py --visual-out v3/fixtures/visual/out; python3 scripts/run_v3_mechanical_smokes.py; python3 scripts/run_v3_feedback_parity_smoke.py; python3 scripts/validate_v3_visual_packages.py v3/fixtures/visual/out/visual-package.json; cargo fmt --all --check; cargo build --workspace --quiet; cargo clippy --workspace -- -D warnings; cargo test --workspace --quiet

Not-tested: Live gauntlet execution; hidden scenario generation; independent reviewer calibration

Co-authored-by: OmX <omx@oh-my-codex.dev>
@graphanov graphanov merged commit 07784d7 into main Jun 3, 2026
1 check passed
@graphanov graphanov deleted the proof/workflow-gauntlet-v1 branch June 3, 2026 15:02
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