| name | semantic-extract |
| description | Extract structured data and entities from existing HTML, Markdown, or text files to audit metadata availability. |
| required_roles | {"scribe":"roles/scribe.editor"} |
| personas | ["content-strategist","data-analyst"] |
Semantic Extractor Skill
Parse documents to identify and extract existing structured data (tables, lists, headers) and potential entities, converting them into a raw structured format for audit or migration.
Inputs
PATH - The source document to process.
OUTPUT_FORMAT - (Optional) "json", "csv", "yaml" (default: "json").
Workflow
- Read: Load content from
PATH.
- Identify:
- Locate HTML tables, definition lists, header hierarchies.
- Detect key-value pairs (e.g., in frontmatter).
- Extract: Pull the raw data from these structures.
- Format: Convert the extracted data into the requested
OUTPUT_FORMAT.
Output
A structured object containing the extracted data, keyed by the source element (e.g., "Table 1", "Frontmatter").