fda-nda

name	fda-nda
description	Fetch recent FDA new approvals from official FDA sources and export them as fdaNDA.xlsx. Use when the user asks for FDA recently approved drugs, biologics, or vaccines within a time window like 6 months or 12 months, mentions fdaNDA, or wants a Chinese summary table of FDA new approvals saved in the current directory.

FDA NDA

Overview

Use this skill to gather recent FDA approvals from official FDA pages only, translate the descriptive fields into Chinese, and save the result as fdaNDA.xlsx in the current working directory.

Keep generic_name and brand_name in English. Translate every other descriptive field into concise Chinese.

The authoritative output is fdaNDA.xlsx. Do not treat fdaNDA.sample.xlsx as the final deliverable unless the user explicitly asks for a sample copy.

Workflow

1. Parse the user input into a positive month count. Accept examples like 12 months, last 6 months, or recent 9 months.

2. Run the bundled end-to-end script.

python "C:\Users\Lenovo\.claude\skills\fda-nda\scripts\run_fda_nda.py" --period "12 months" --output-dir "." --keep-json

3. The end-to-end script already:

   • collects raw FDA records
   • converts them into Chinese summary rows
   • writes .\fdaNDA.xlsx
   • optionally saves .\fdaNDA.raw.json and .\fdaNDA.curated.json

4. The collector stage filters by date and merges:

   • CDER annual novel-drug approvals pages
   • CDER annual report PDF flags
   • openFDA sponsor/application metadata
   • CBER biologics approval pages
   • linked CBER product pages

5. The final workbook rows must include these fields:

   • drug_type_zh
   • generic_name
   • brand_name
   • manufacturer_zh
   • indication_zh
   • approval_date
   • significance_zh
   • source_urls

   If manufacturer_en is blank, do one manual FDA-only supplement pass before finalizing:

   • search official FDA pages for the product name
   • prefer an FDA approval letter, FDA product page, or Drugs@FDA/openFDA result
   • then fill manufacturer_zh

6. Confirm .\fdaNDA.xlsx exists and tell the user where it was written.

Translation Rules

• Keep generic_name and brand_name unchanged.
• Translate manufacturer_zh, indication_zh, and significance_zh into Chinese.
• For manufacturer cells, a Chinese wrapper around the official company name is acceptable.
• Keep approval_date in ISO format YYYY-MM-DD unless the user asks for another date format.
• If manufacturer_en is missing after the collector step, fill it from another official FDA page before exporting.

Significance Rules

Build significance_zh only from the official evidence already present in the raw JSON.

• Prefer official_evidence_en.report_highlight when it exists.
• Otherwise use regulatory_flags_en and modality_hint_en.
• Do not invent head-to-head superiority claims.
• If the evidence only shows regulatory importance, say so explicitly.

Good patterns:

• State that FDA identified a product as first-in-class when the raw flags say so.
• State that a product used breakthrough, priority, or accelerated review when the raw flags say so.
• State that a product is notable for being a gene therapy, biologic platform, or vaccine platform when the raw evidence supports it.

Avoid:

• claims like "best efficacy"
• claims like "better than all competitors"
• any comparison not supported by the raw FDA evidence

If The Collector Returns Warnings

• Read fda_sources.md only if you need to verify source coverage.
• If a current-year FDA page is unavailable, tell the user the workbook reflects all currently reachable official pages returned by the script.
• Still export the workbook if valid records were collected.

Output Standard

The final workbook should include Chinese column headers and Chinese descriptions for all non-name fields.