[deep-report] Close Claude vs Copilot smoke-test/example parity gap (9 vs 15 variants, 58 vs 115 engine declarations) #65084

Description

engine: claude (or nested id: claude) appears 58 times across .github/workflows/*.md versus 115 for Copilot (~1.98x), and Claude-specific smoke-test variants number only 9 (smoke-claude*.md, smoke-github-claude.md, smoke-claude-on-copilot.md) versus 15 Copilot variants (arm64, SDK, sub-agents, AOAI apikey/Entra auth modes, small-model, etc). A Claude-only user has roughly half the reference material to copy from when building a new workflow, and several auth/runtime modes Copilot covers (sub-agents, ARM64, alternate auth) have no Claude equivalent smoke test.

Expected Impact

Closes the reference-material gap for Claude users and gives CI equivalent coverage of Claude-specific runtime modes (ARM64, SDK-style invocation, sub-agents, alternate auth) that currently only Copilot smoke tests exercise.

Suggested Agent

GitHub Copilot coding agent or a workflow-authoring agent — add 4-6 new smoke-claude-*.md workflows mirroring the Copilot variants (e.g. smoke-claude-arm64.md, smoke-claude-sdk.md, smoke-claude-sub-agents.md, an alternate-auth-mode variant), following the existing smoke-claude.md / smoke-copilot-*.md structure, then run make recompile.

Estimated Effort

Medium (1-4 hours)

Data Source

DeepReport Intelligence Briefing — 2026-10-02, sourced from Claude Code User Documentation Review #65018 ("Major Obstacles" — example/smoke-test parity gap), a pattern flagged in every run since at least 2026-08-22 but not previously filed as a standalone issue.

Generated by 🔬 Deep Report · claude · agent · 212 AIC · ⌖ 7.99 AIC · ⊞ 7.3K · ◷

• expires on Oct 4, 2026, 10:42 AM UTC-08:00