[deep-report] engine-example-counter undercounts nested 'engine: {id: ...}' form, making engine-parity trend data unreliable #55132

Description

claude-code-user-docs-review's engine-example-counter sub-agent (.github/workflows/claude-code-user-docs-review.md:168-180) counts engine usage across example workflows by scanning .github/workflows/*.md for literal occurrences of engine: claude, engine: copilot, engine: codex, engine: custom. This misses the nested-mapping form of the engine: field (engine:\n id: copilot\n ...), which is common in the repo — e.g. .github/workflows/delight.md declares engine:\n id: copilot\n copilot-sdk: true.

Verified live counts in this repo:

• Literal engine: claude / engine: copilot (single-line form): 37 / 34 files
• Nested engine:\n id: claude / engine:\n id: copilot form: 24 / 101 additional files

The literal-string counter therefore undercounts Copilot examples by roughly 3x and Claude examples by ~65%, missing the majority of nested-form declarations entirely. This directly explains the "Trend note" flagged by the workflow's own most recent report (discussion #55075): "engine-example counts shifted sharply since 2026-08-22, claude 39→60, copilot 39→138 — worth confirming whether this reflects real example growth or a counting-methodology change." It's the latter — the count is unreliable and swings based on which YAML form authors happen to use, not real example growth/decline.

Proposed Fix

Update the engine-example-counter agent's counting logic to also match the nested engine:\n id: <name> form (e.g. via a YAML-aware parse or a regex covering both forms), so engine-example counts and the Copilot-vs-Claude parity trend this workflow reports on are accurate and stable across runs.

Expected Impact

Fixes a data-quality bug in a recurring report — the "Engine & Tool Matrix" and parity trend note are currently misleading rather than informative.

Suggested Agent

general-purpose — fix the counting logic in the engine-example-counter sub-agent definition.

Estimated Effort

Quick (< 1 hour)

Data Source

DeepReport Intelligence analysis, discussion #55075 ([claude-code-user-docs-review] Claude Code User Documentation Review — 2026-08-23) "Trend note". Verified live via grep counts against .github/workflows/*.md.

Generated by 🔬 Deep Report · agent · 212.6 AIC · ⌖ 15.7 AIC · ⊞ 12.4K · ◷

• expires on Aug 25, 2026, 10:35 AM UTC-08:00