[Schema Consistency] Schema consistency check — 2026-11-10

[github-actions[bot]]

🔍 Schema consistency check — 2026-11-10

Three actionable inconsistencies were found across schema, parser, and documentation. The precomputed workflow scan found no top-level keys missing from the schema. Strategy: cached field-gap triage and targeted constraint/default comparisons (reused; high effectiveness).

Key issues

1. Zero limits accepted but discarded. The schema defines max-turns and max-tool-denials using the non-negative templatable_integer (schema:4956, schema:15601). Both parser functions require at least 1 and return an empty value for zero (engine_config_parser.go:56, engine_config_parser.go:121); tests explicitly cover this behavior. A schema-valid max-tool-denials: 0 is treated as unset, so the Copilot SDK default of 5 applies; max-turns: 0 similarly does not set the requested limit. Change both fields to the positive-integer schema, or deliberately implement and document zero semantics, and add schema/compiler regression coverage.

2. Daily credit default described incorrectly. The schema says omitting max-daily-ai-credits disables the guardrail (schema:4984). The normal resolution path instead falls back to the repository variable or built-in 5000 (daily_aic_workflow.go:207, constants.go:443); the cost reference agrees (cost-management.md:287). Correct the schema description and regenerate derived reference text; document applicable special execution paths separately. Incorrect guidance can lead authors to assume the guardrail is off.

3. Generated reference contains invalid examples. The full frontmatter reference uses example-value as the expression example for max-turns, max-tool-denials, and max-tool-calls, and as a credit-limit string (frontmatter-full.md:4325). These strings match neither the schema expression pattern nor the numeric credit patterns (schema:15601, schema:15647). Generate valid examples such as ${{ inputs.max-turns }} and 1M, and validate generated examples against their field schemas.

Scope and evidence

The field-gap scan flagged docker-image-pull-policy, skip-if-decision, and work-queue-policy as absent from the typed frontmatter structure. Targeted review found each is handled through raw frontmatter, so these gaps alone are not defects. The precomputed scan reported no top-level workflow keys absent from the schema; targeted workflow searches also found normal positive examples of the reviewed limits. This was a focused review, not exhaustive validation of every workflow or nested field. No AWF canonical-source comparison was used.

Strategy performance: 3 findings; reuse the cached method, including checks for deliberate raw-frontmatter handling and generated examples.

Next actions

• Align the two zero-limit schema and parser contracts, with regression tests.
• Correct the daily-credit default in the schema and generated documentation.
• Emit and validate valid generated examples.

References: §42761852081

Generated by ✅ Schema Consistency Checker · codex · routed: gpt61 medium · ~9 AIC · ⌖ ~2 AIC · ⊞ 470 · ◷

• expires on Nov 10, 2026, 9:29 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Schema Consistency Checker.

A newer discussion is available at Discussion #76770.