You are running a source-grounded MadCowork Office E2E workflow.
Communicate final user-facing conclusions in Traditional Chinese, but all tool planning and tool arguments must be concise English.
Topic: Samsung Electronics 2026 memory-cycle analysis: Samsung memory/semiconductor results, DRAM/NAND pricing, inventory turnover days, and AI infrastructure insight.
Search and fetch evidence yourself. Start with these source targets and search terms, but verify fetchability:
- Samsung Electronics investor relations earnings release page: https://www.samsung.com/global/ir/financial-information/earnings-release/
- Search: Samsung Electronics Q1 2026 results memory business revenue operating profit
- Search: Samsung Electronics FY2025 results Device Solutions memory business
- Search: Samsung Electronics Q1 2025 results memory semiconductor
- TrendForce DRAM price page: https://www.trendforce.com/price
- StockAnalysis Samsung Electronics income statement: https://stockanalysis.com/quote/krx/005930/financials/
- StockAnalysis Samsung Electronics balance sheet: https://stockanalysis.com/quote/krx/005930/financials/balance-sheet/
Hard requirements:
1. Do not rely on model memory. Use web_search, web_fetch, and web_crawl before making factual or numerical claims.
2. Gather evidence from multiple sources, including Samsung official IR/newsroom reports and at least two third-party market or financial data sources.
3. Create CSV files in this session workspace:
- samsung_sources.csv with columns: source_type,title,url,published_date,used_for.
- samsung_financials.csv with columns: company,metric,period,value,unit,source_url,note.
- memory_prices.csv with columns: product,price_type,period,value,unit,source_url,note.
- inventory_metrics.csv with columns: company,metric,period,value,unit,formula,source_url,note.
- samsung_chart_inputs.csv with columns: chart_name,series,period,value,unit,source_url,note.
- session_metadata.csv with columns: artifact,type,path,created_at,source_inputs.
4. Every numeric claim used in the deck must have a source_url in one of the CSV files.
5. Include these analyses if sources support them:
- Samsung Electronics Q1 2026 revenue and operating profit.
- Device Solutions / semiconductor / memory business revenue and operating profit for Q1 2026 or latest available quarter.
- FY2025 revenue, operating profit, net income, and operating margin.
- FY2025 cost of revenue / cost of sales and inventory from a financial-data source or official financial statements.
- Annual inventory days / DIO using: average inventory / cost of revenue * 365. Use FY2025 and FY2024 inventory if available. If only one period is available, clearly label the limitation.
- DRAM, DDR4/DDR5, NAND, or HBM price / price-change data from TrendForce or another fetchable third-party memory market source.
- HBM / AI memory demand insight only when backed by a fetched source.
6. Create chart-ready rows in samsung_chart_inputs.csv for revenue/profitability, segment mix, inventory/DIO, memory pricing, and sourced AI-memory insight.
7. Call source_validate on samsung_sources.csv with min_official_sources=2 and min_third_party_sources=2. If market data cannot be fetched, record the data gap in the deck instead of inventing it.
8. Before charting or xdeck, call list_skills, read xdata-office-deck-workflow and xdata-spreadsheet-workflow if available, call csv_profile on the financial/chart CSVs, then call csv_validate on every CSV.
If an evidence file contains Markdown tables, call extract_markdown_tables first and use the extracted CSV tables as the source for numeric extraction.
9. Generate a self-contained web presentation through xdeck_generate. Set out_dir exactly to "samsung-memory-web-deck". Pass facts_paths, sources_path, chart_data_path, and validation_report_path when available.
10. The deck should be 8 slides, Traditional Chinese, investor / memory-cycle strategy briefing style, with speaker_note on each slide.
11. Run xdeck_audit and xdeck_screenshots. If the audit finds layout issues, run one xdeck_revise pass.
12. Final answer: list generated artifact paths and briefly explain the evidence chain.
Important: if a source cannot be fetched, record the failed URL in samsung_sources.csv with used_for="attempted but unavailable" and continue with fetchable sources.
Automatic correction pass for the Samsung memory E2E workflow.
The harness detected missing/non-numeric DIO, memory-price values, or chart inputs. Do not rely on model memory. Read the saved evidence files in this session workspace and repair the CSVs from evidence only.
Required corrections:
1. List evidence/ and read the saved Samsung official evidence files, StockAnalysis income statement evidence, StockAnalysis balance sheet evidence, and TrendForce price evidence.
Read each evidence file at most once. After reading it, extract the needed rows from the content already in context and move on to write_file/csv_validate.
Prefer the already-saved HTML evidence files for correction. If a source is a PDF, use read_pdf; do not use read_file on a PDF or a missing downloaded PDF path.
For StockAnalysis and TrendForce evidence, call extract_markdown_tables and use the generated CSV table paths before repairing the final CSVs.
2. Update samsung_financials.csv with missing Cost of Revenue / Cost of Sales rows from the financial statement source for FY2025 and FY2024 if available.
3. Update inventory_metrics.csv with FY2025 inventory, FY2024 inventory if available, FY2025 average inventory, FY2025 cost of revenue, and FY2025 DIO. Calculate DIO as average inventory / cost of revenue * 365. Values must be numeric, not N/A or unknown.
4. Update memory_prices.csv with numeric TrendForce rows. Use session average and percentage change rows when available. Qualitative HBM/NAND rows are allowed only if they are not used as numeric price chart inputs.
5. Update samsung_chart_inputs.csv so Inventory / DIO and Memory Pricing charts contain numeric values, not N/A, unknown, trend strings, or qualitative words.
6. Run csv_profile on samsung_financials.csv, inventory_metrics.csv, memory_prices.csv, and samsung_chart_inputs.csv. Run csv_validate on every CSV and source_validate on samsung_sources.csv.
7. Regenerate the same 8-slide web presentation by calling xdeck_generate with out_dir exactly "samsung-memory-web-deck", passing the updated CSVs. Then run xdeck_audit. If audit has findings, run xdeck_revise once, then run xdeck_screenshots.
8. Final answer in Traditional Chinese: explain that this was an automatic correction pass fixing inventory/DIO and numeric memory-price extraction.
