semantic_llm_assessments

semantic_llm_assessments(x)

Target-level semantic assessments attached to a dictionary or a package.

The companion accessor to :func:semantic_suggestions, for the semantic_llm_assessments attribute that :func:~metasalmonpy.ingest_semantic_assessments (and the deprecated suggest_semantics(llm_assess=True)) attaches. Reading assessments is never itself a model call: this only reports assessments that already exist.

For a package path, the record :func:~metasalmonpy.ingest_semantic_assessments persisted in review/semantic-llm-assessments.csv, typed as the 30-column assessment row and carrying the validator findings column by column in attrs["semantic_validator_findings"] (pd.DataFrame() of it is the findings frame; it is not the frame itself, so that two records can be concatenated), or None when no record has been ingested. The in-package model call never writes one. A review/ directory or record that is a symbolic link is refused rather than followed.