Cut publication commands over to arena

[vaibhav-i]

Summary

• delegate competitive, publication-bundle, and export-publication from ax-eval to the arena CLI through the shell-free allowlisted compatibility launcher
• remove duplicate core publication/export and cross-target competitive-rendering implementations
• keep competitive reporting restricted to --from <sealed-publication-bundle>; legacy raw --results input fails closed
• accept legacy publication selectors only when suite, vendor, and effort choices exactly match the immutable batch
• add import guards preventing arena publication policy or competitive rendering from returning to core

Trust and compatibility

Official bundle, export, and competitive outputs remain bound to complete pinned-oci + hosted-trusted production evidence with a verified detached GitHub OIDC attestation. Missing, incomplete, local, unsealed, or invalid inputs emit no bundle. The deprecated ax-eval names remain one-minor process-launcher aliases, but they cannot bypass the sealed arena path.

Verification

• core tests: 549 passed
• arena tests: 207 passed (--maxWorkers=1 locally to avoid host-load timeout contention); focused cutover tests: 21 passed at default concurrency
• npm run typecheck
• npm run build
• npm run pack:check
• core and arena npm --cache .npm-cache pack --dry-run
• import-boundary checks and git diff --check

All tests are offline and keyless. No live or credentialed evaluation was run.

Independent audit

A separate review-only agent audited exact commit 1327479 against PR #200. It found two documentation inconsistencies; both were fixed, then the amended commit was re-audited. Final audit: clean with no actionable security, compatibility, ownership, or correctness findings.

Stack

Base: #200 core DAEB execution-policy removal.

Rollback: revert this PR to restore the duplicate core publication and competitive implementations while leaving arena implementations intact.

Next slice: move DAEB pack-composition policy into arena-owned authoring APIs.

[github-actions]

Normalized records diff

Overall is a macro-average of participating surface scores. Agents are never averaged together.

Overall by agent

Product	Agent	Base pass@1	Head pass@1	Δ	Base pass³	Head pass³	Surfaces
fixture-db	claude-code	92.9%	92.9%	+0.0 pp	78.6% (11/14)	78.6% (11/14)	2
fixture-db	codex	78.6%	78.6%	+0.0 pp	64.3% (9/14)	64.3% (9/14)	2

Surface cells

Product	Agent	Surface	Base pass@1	Head pass@1	Δ	Base pass³	Head pass³	Latency ms	Cost USD	Tokens	Harness version
fixture-db	claude-code	api	85.7%	85.7%	+0.0 pp	71.4% (5/7)	71.4% (5/7)	1100 → 1100	0.4200 → 0.4200	3350 → 3350	2.1.5 → 2.1.5
fixture-db	claude-code	cli	100.0%	100.0%	+0.0 pp	85.7% (6/7)	85.7% (6/7)	1300 → 1300	0.5100 → 0.5100	3720 → 3720	2.1.5 → 2.1.5
fixture-db	codex	api	85.7%	85.7%	+0.0 pp	71.4% (5/7)	71.4% (5/7)	1200 → 1200	— → —	3600 → 3600	0.121.0 → 0.121.0
fixture-db	codex	cli	71.4%	71.4%	+0.0 pp	57.1% (4/7)	57.1% (4/7)	1500 → 1500	— → —	4100 → 4100	0.121.0 → 0.121.0

Operational metrics are context only and never affect ranking.