P233 · Evaluation & feedback

Validate the consumer interpretation

A local lenient reader passing is not proof that the real consumer accepts the same artifact.

Editorially reviewed

These examples and illustrative results are independently authored teaching materials, not measured model results.

Use case

A skill description contains colon-space: Use this: when X. A lenient split-on-first-colon helper reads a string, but that does not establish acceptance of full frontmatter by a YAML consumer.

Mechanism

Identify consumer, format contract and parser version. Validate the full serialized artifact rather than the helper’s parsed object; encode scalars correctly where quoting is needed. Check valid and invalid neighbors so the validator’s rejection behavior is known, preserving field type/meaning. Test actual loading when available; disclose when only a strict reference parser was used.

Bad example

A local helper read the description, so publish the frontmatter without checking the consumer grammar.

Good example

Validate complete frontmatter against the target grammar, quote the description correctly and check its parsed string value. Test unquoted colon-space and unclosed-quote neighbors. Record parser and loading scope rather than treating lenient success as compatibility.

Why the change matters

Producer and consumer must interpret the same bytes. Object-level success can hide serialization defects; neighboring invalid examples test the validator’s rejection ability.

Observable expectation

Teaching valid fragment:

description: "Use this: when X"

Its parsed value retains the full string. Unclosed quotes fail, and unquoted colon-space is rejected under strict grammar. The cited narrow checker covers selected shapes, not full YAML conformance.

Limits

Strict parsing does not establish host discovery, loading or execution; semantics need the consumer contract. Parser and host restrictions differ. Frozen historical host failures are not facts about every current product; verify actual versions.

Sources and evidence

Read the editorial criteria