Ghostbill Backend V2

Use this skill for Ghostbill backend work that must preserve legacy CSV behavior while adding or maintaining XLSX, JSON, and PDF support in the existing pipeline.

Workflow

1. Read references/source-prompt.md in full before editing. It is the authoritative specification.
2. Detect whether a legacy CsvParsingService already exists. If it does, treat it as frozen. If it does not, create the baseline exactly as required before any non-CSV parser work.
3. Implement in the sequence defined by the reference prompt: contracts and analysis first, then CSV validation, then XLSX/JSON, then PDF strategies, then controller and DI wiring.
4. Treat all exact backend contracts as fixed: file paths, namespaces, interfaces, parser precedence, alias lists, regexes, constants, error codes, and output ordering.
5. Keep parsers translation-only and format-agnostic from the analysis layer.
6. Run the required build and test verification before finishing.

Guardrails

• Do not change CsvParsingService internals, signature, or observable output when a legacy implementation exists.
• Do not alter the API route, request fields, DTO shapes, pipeline semantics, or analysis thresholds unless explicitly instructed outside this skill.
• Do not add OCR, nondeterministic behavior, or business logic inside parsers or shared parsing helpers.