How To Extract 10K Mda Quotes From Pasted Text Only
MD&A memos go bad when a model fetches EDGAR and invents a footnote. You pasted Item 7. The extractor should quote that paste, keep units, and refuse a buy/sell line.
The matching generator is the 10-K MD&A Extractor that Quotes Only Pasted Text prompt. Browse related cards in the PromptDig library (Browse more prompts). When a filled run survives, share the version you actually use (Share a prompt).
Banner not-advice and no fetch
Extract MD&A themes from a pasted 10-K excerpt. Quote only pasted text. No invented figures, tickers, or footnote citations. Start by filling Inputs, not by asking the model to remember last week's run. If a field is blank, write NONE or NOT IN INPUTS and leave it blank through Generate. The card is built so the model cannot honestly invent a number, owner, URL, or command that you did not paste.
Paste these fields before you hit run:
Issuer name and ticker as they appear in the paste: [Issuer]
Fiscal period the excerpt covers: [Period]
Pasted MD&A or Item 7 text: [Excerpt]
Questions I want answered: [Questions]
Units I must keep (USD thousands, etc.): [Units]
Items I must not infer: [NoInfer]
Words I must not use: [Banned]
Max quotes: [QuoteCap]
Output format: [Format]
Not-advice line: [Disclaimer]
That inventory is the honesty ledger. Anything that does not appear there is forbidden in the draft. If you catch yourself adding a nice-to-have after the run, you are no longer using the card. You are ghostwriting. Put the extra fact in Inputs and run again.
Ledger every number in the excerpt
Generate is numbered on purpose. Do not skip a step because the first paragraph looked done. The early steps exist to stop later prose from smuggling claims.
Walk the Generate list in order:
- Banner: not investment advice. Analysis of pasted text only. No EDGAR fetch.
- Honesty ledger: issuer, ticker, period, every number and named risk in Excerpt. Forbidden: numbers not in Excerpt.
- Theme list: only themes that have a quote. Each quote is a verbatim fragment plus a locator (paragraph or heading in the paste).
- Question grid: each Questions row, answer with a quote or NOT IN EXCERPT.
- Units: keep Units. Do not rescale millions to billions unless Excerpt does.
- NoInfer: list attempted inferences you refused.
- Gaps: headings Item 7 usually has that this excerpt omitted.
- Compliance pass: quote Banned words, invented figures, and extra tickers. Cut them. Honor QuoteCap.
If a step asks for a version lock, quote the version from Inputs in the output. If a step asks for a refuse list, keep the refuse list in the published artifact, not in a sidebar you delete. Reviewers should see what the model was not allowed to do.
Answer questions with a quote or NOT IN EXCERPT
Most failures are the same shape: a missing field gets a confident fill. A conversion rate appears. A Gradle task appears. A flash point appears. A caption appears on a job that asked for slide text only. Your review is to search the draft for numbers, names, and commands, then grep Inputs. No match means cut.
Honor the constraints as hard stops, not vibes:
- Quote only pasted text. No web lookup.
- Never invent a financial figure or footnote.
- Not investment advice. No buy/sell.
- If ticker is missing from the paste, do not guess it.
- Stay at or under QuoteCap quotes.
When the card says not legal advice, not certification, not an exam dump, or not a caption engine, that sentence belongs at the top of the output. Deleting it to look more finished is how you inherit risk.
Do not compute ratios you were told not to infer
Finish with the compliance pass the prompt already asks for. Quote the banned-word hits. Cut them. Print character counts when the job has a cap. Print word counts when the job has a budget. List gaps as gaps. Five missing facts are more useful than one smooth paragraph.
Tags on the card (10-k mda extractor, sec 10-k quote only, mda no invented figures) are a reminder of the job shape, not an invitation to wander into a neighboring cluster. If you need a different surface, open a different PromptDig card rather than stretching this one.
Fill the card, then run
Replace every bracket. Run on ChatGPT, Claude, or Gemini. Read the ledger first, then the artifact. If the model invents a commit, KPI, DOI, PEL, bid, or logo, discard the run. Tighten Inputs. Run again. Share the filled card that survived, not the first draft that sounded done.