fix: expand us-cpa extraction review and rendering

This commit is contained in:
Stefano Fiorini
2026-03-15 03:01:16 -05:00
parent 6c02e0b7c6
commit fb39fe76cb
19 changed files with 693 additions and 56 deletions

View File

@@ -46,7 +46,8 @@ skills/us-cpa/scripts/us-cpa extract-docs --tax-year 2025 --case-dir ~/tax-cases
- IRS materials first; escalate to primary law only when needed
- stop on conflicting facts and ask the user to resolve the issue before continuing
- official IRS PDFs are the target compiled-form artifacts
- overlay-rendered forms must be flagged for human review
- deterministic field-fill is the preferred render path when the official PDF exposes usable fields
- overlay-rendered forms are the fallback and must be flagged for human review
## Output
@@ -55,10 +56,11 @@ skills/us-cpa/scripts/us-cpa extract-docs --tax-year 2025 --case-dir ~/tax-cases
- `question` supports `--style conversation|memo`
- `fetch-year` downloads the bootstrap IRS form/instruction corpus into `~/.cache/us-cpa` by default
- override the cache root with `US_CPA_CACHE_DIR` when you need an isolated run or fixture generation
- `extract-docs` creates or opens a case, registers documents, stores facts, and stops with a structured issue if facts conflict
- rendered form artifacts are currently generated by overlaying values onto the official IRS PDFs and are flagged for human review
- `prepare` computes the supported 1040 subset and writes normalized return/artifact/report files into the case directory
- `extract-docs` creates or opens a case, registers documents, stores facts, extracts machine-usable facts from JSON/text/PDF sources where possible, and stops with a structured issue if facts conflict
- `question` currently has explicit IRS-first answers for standard deduction, Schedule C, Schedule D, and Schedule E questions; other questions escalate to primary-law research with official IRC/regulation URLs
- rendered form artifacts prefer fillable-field output when possible and otherwise fall back to overlay output
- `prepare` computes the current supported federal 1040 package, preserves fact provenance in the normalized return, and writes normalized return/artifact/report files into the case directory
- `export-efile-ready` writes a draft transmission-ready payload without transmitting anything
- `review` recomputes the return from case facts, checks artifacts, and returns findings-first output in conversation or memo style
- `review` recomputes the return from case facts, checks artifacts, flags source-fact mismatches and likely omissions, and returns findings-first output in conversation or memo style
For operator details, limitations, and the planned case structure, see `docs/us-cpa.md`.