Question Set Review

April 28, 2026 · View on GitHub

Author: subagent-driven authoring (CGBench v1 Plan 2 Tasks 4-9) Date: 2026-04-28

Counts

TaskFileCountSpec targetNotes
Atask-a.jsonl1260NL→code, 3/lang × 4
Btask-b.jsonl1248structural (callers/extenders/importers), 3/lang × 4
Ctask-c.jsonl832multi-hop (≤3 hops), 2/lang × 4
Dtask-d.jsonl830bitemporal (5 point-in-time, 3 range)
Etask-e.jsonl830linked code+knowledge, 2/corpus × 4
Ftask-f.jsonl1030doc ingestion, 2/format × 5

Total: 58 questions (vs spec target 230). v0.1 sample.

Authoring discipline applied

  • Every gold ID was verified against the actual corpus by git grep or by reading frontmatter.
  • NL phrasings for Task A describe behavior, not paraphrased docstrings.
  • Multi-gold questions include all acceptable answers, not just one.
  • Task C hop distances were hand-traced through the call graph; capped at 1-2 hops in v0.1 (3-hop deferred for clarity).
  • Task D supersession queries reference the knowledge-003 → knowledge-009 pair authored in Task 1.
  • Task E linkage was verified by checking that knowledge-doc references: frontmatter contains the gold code symbols.
  • Task F: each fact file is tested by exactly one question; format-tagging is informational metadata pending Plan 4's per-format ingestion gating.

Smoke results (lexical adapter, Plan 1's CodeGraph baseline)

TaskPer-language MRR / Recall@10 (where applicable)
APy 0.23, TS 0.83, Go 0.07, Rust 1.0 (MRR)
BPy 0, TS 0, Go 0.083, Rust 0 (R@10)

The low scores on Task B confirm the lexical adapter's structural-retrieval gap — the benchmark exercises CodeGraph's actual differentiator (graph edges) only when Plan 4 swaps to a graph-aware query path.

Known gaps for the spec's full counts

  • Per-language coverage at v0.1 is 2-3 questions per language vs spec's 8-15. Statistical confidence is correspondingly lower.
  • No second labeler reviewed this set. Plan 4 should commission one before publishing.
  • Task F's per-format ingestion gating is deferred (the format field is metadata only in v0.1).
  • PDF rendering disabled (no pdflatex on author's machine; install basictex for Plan 4 publishing).

Reviewer sign-off

  • Self-review pass complete (this document)
  • Schema-validation tests pass
  • Independent reviewer pass (deferred to Plan 4)