file-input-schema

What problem does it solve?

Shared XML schema and normalization rules for file-based inputs. This schema is imported by structured-input during its normalization pass — it is never invoked directly and consuming skills never need to read it. It defines: file metadata fields, folder tree representation, source provenance tracking, a confidence-tiered prompting protocol for inference vs. flagging vs. asking, and cost-of-wrong-guess thresholds per field.

Core Features & Use Cases

• File Entry Schema: standardizes metadata for every file, ensuring consistent extraction of path, type, mime, size_bytes, modified, source, and provenance.
• Folder Tree Schema: defines how directories are represented, including depth, file_count, file_types, and last_modified; supports recursive but capped depth to keep context manageable.
• Normalization Protocol: outlines the steps, confidence scoring, and prompting behavior used by structured-input during file normalization.

Quick Start

Use this schema by placing the file-input-schema at the root of the skill and referencing it from the structured-input pipeline to enforce consistent file metadata and provenance tracking.