Browse Prompts
1018 prompts available · Page 75 of 85
RFP Extract to Compliance Matrix and NO_DATA Draft
Extract RFP shall-statements into a compliance matrix and a NO_DATA draft so you do not hallucinate answers.
Act as a proposal operations lead. Extract an RFP into a compliance matrix, then draft only where the user's facts exist. Use NO_DATA instead of guessing. This is not a pricing one-pager and not a full grant narrative. Inputs: - RFP text or section paste: [Paste] - Our facts we can prove: [Facts] - What we will not bid: [Out] - Page or word limits: [Limits] - Required headings: [Headings] - Past performance I may name: [Past, with permission] Generate: 1. Shall/should inventory: quote each requirement, page/section if present, MUST/SHOULD. 2. Compliance matrix: Req ID, quote, response (COMPLY / PARTIAL / NO_DATA / OUT), fact pointer, owner blank. 3. Conflict list: requirements that fight each other, or fight Out. 4. Draft: only COMPLY rows, under Limits, with Facts in-line. PARTIAL gets a question, not a fake yes. 5. NO_DATA pack: a list of questions to the capture lead. Do not fill with industry filler. 6. Past performance: only from Past. If none, skip. Constraints: - Never invent a certification, SOC 2, or insurance limit. - Do not write a price. - Quote the RFP; do not paraphrase away a shall.
Grant Proposal Narrative with No Invented Stats
Draft a funder-aligned grant narrative from your notes with SMART objectives and NO_DATA where numbers are missing.
Act as a grant writer. Draft a narrative from the user's notes. Do not invent beneficiaries, outcomes, or dollars. Align to the funder language they pasted. This is not a press release or a case study. Inputs: - Funder and opportunity: [Funder] - Word or page cap: [Cap] - Need statement notes: [Need] - What we already do: [Program] - Numbers I can prove (with source): [Proof] - Budget story I can tell: [Budget] - Partners who have agreed: [Partners] - What I must not claim: [Banned] Generate: 1. Gap list: every number in Need that is not in Proof. Mark NO_DATA. Do not fill with national averages unless Proof includes them. 2. Outline mapped to the funder's requested headings. If they did not paste headings, use Need, Approach, Objectives, Capacity, Budget story. 3. SMART objectives (3): only from Proof or from program design that does not require a fake baseline. If baseline is missing, write NEED BASELINE. 4. Narrative draft within Cap. Cite Proof in-line as (source, year). No fake evaluations. 5. Budget story: dollars only from Budget. No padding. 6. Refusal log: invented stats, unnamed celebrities, guaranteed outcomes. Constraints: - No invented census tracts or disease rates. - Partners only if they have agreed in Inputs. - Not a full application package (no forms).
Fair Housing MLS Listing Description (Agent-Side)
Write an agent-side MLS description with character counts and a Fair Housing banned-phrase pass, distinct from a tenant rental listing.
Act as a listing copywriter for an MLS agent remarks field. This is agent-side sale or lease listing copy, not the Honest Rental Listing writer for a landlord's Craigslist post. Enforce Fair Housing. Do not invent square footage, schools, crime, or "safe" neighborhoods. Inputs: - Property type and address area: [Type, city, no steering] - Facts from the data sheet: [Beds, baths, sqft if measured, HOA, year] - What I personally verified: [Verified] - What I have not verified: [Unverified] - MLS character cap: [Public remarks cap] - Brokerage compliance notes: [Notes] - Photos I can truthfully describe: [Photos] Generate: 1. Banned-phrase check: list Fair Housing risk words I will not use (family-friendly, perfect for singles, walking distance to church, safe, exclusive, able-bodied, no kids, great for empty nesters, coded race/religion/national origin). Scan Inputs and quote any hits. 2. Public remarks: within cap, print count. Facts only. No school ratings. No crime. No buyer profile. 3. Agent-only remarks: logistics (lockbox, occupancy) without steering. 4. Feature bullets (6) tied to Verified. Unverified gets "ASSESSOR / UNVERIFIED" or is dropped. 5. Photo captions (5) matching Photos. No people as selling points. 6. Compliance rewrite: if I had to cut a line, show before/after. Constraints: - Fair Housing is a hard pass. When unsure, cut. - Distinct from tenant-facing rental ads. - No invented sqft or tax amounts.
Tool-Call Agent Prompt-to-Eval Rubric
Turn an agent prompt into an eval rubric with tool-call gold traces, traps, and a pass/fail rule. No fake benchmark scores.
Act as an eval engineer. Turn an agent system prompt into a rubric plus cases. Distinct from the generic Prompt A/B harness and from RAG eval sets. Focus on tool calls: when to call, which args, when to refuse. Inputs: - Agent prompt: [Paste] - Tools (names + args): [Tools] - Failures we already saw: [Fails] - Must-not: [PII, payments, live prod] - Budget: [How many cases] Generate: 1. Behaviors to score: tool choice, args, refusal, final answer groundedness. 0-2 scale each. 2. Gold traces: for each case, the tool sequence (or none). 3. Cases: Budget count, mix of happy path, trap (should not call), missing-arg, tool error. 4. Decision rule: pass if mean meets a threshold you state. Fail the run on any must-not hit. Do not invent a public leaderboard score. 5. Trap ideas from Fails only plus obvious PII. 6. Refused: fabricating tool JSON the schema does not have. Constraints: - If Tools is empty, say this prompt cannot be tool-eval'd yet. - No fake 92 percent accuracy.
GitHub Copilot instructions.md Generator
Generate a GitHub Copilot instructions.md for a repo with stack, never-list, and test commands. Distinct from Cursor project rules.
Act as a staff engineer writing GitHub Copilot instructions.md. Distinct from Cursor project-rule files. Do not invent runtime versions. Inputs: - Repo purpose: [Repo] - Stack I can prove: [Stack] - Commands that work: [Commands] - Never-list: [Never] - Paths that need extra care: [Paths] - Secrets: [Secrets policy] Generate: 1. Map: Copilot instructions.md vs path notes. Do not emit Cursor mdc. 2. Full markdown: purpose, commands, style, never-list, secrets. 3. Path notes as headings, not YAML front matter. 4. Install: where the file goes. Humans fill versions. 5. Refused: print env files, disable safety, invent Node version. Constraints: - No Cursor globs YAML. - No fake CI badges. - Keep under 400 lines.
Support Agent System Prompt with Escalation and PII Rules
Write a support-agent system prompt with identity, scope, escalation, and PII/PHI rules, distinct from a reply playbook.
Act as a conversation-design lead. Write a system prompt for a support agent LLM. Distinct from the Customer Support Reply Playbook (that is a human reply pack). This is identity, tools, escalation, PII. Inputs: - Brand and channels: [Brand] - What the bot may do: [Scope] - Tools: [order lookup, none] - Escalation: [When, to whom] - PII/PHI rules: [What not to collect or echo] - Tone: [Tone] - Banned claims: [Banned] Generate: 1. System prompt (paste-ready): Identity, Scope, Style, Tools, Escalation, PII, Never-list, Closing. 2. Example turns (3): in-scope, escalate, PII refuse. 3. Tool-call sketch: when to call order lookup, what to do if tool fails. No fake APIs. 4. Distinct from playbook: do not write 10 canned emails. 5. Refused: asking for SSN, full card number, medical diagnosis. Constraints: - If Tools is none, the bot must not pretend to have looked up an order. - Do not leak other customers. - Keep the system prompt under 800 words.
Gemini Gem Instructions: Role, Goal, Input, Output
Write Gemini Gem instructions in Role / Goal / Input / Output / Constraints form with starters and a never-list.
Act as a Gemini Gem author. Output a Gem using Role, Goal, Input, Output, Constraints. Distinct from Custom GPT instructions and from Claude Project instructions. Inputs: - Gem name and job: [Job] - User: [User] - Gems tools they enabled: [Drive, browsing, none] - Never-list: [Never] - Examples they want inside the Gem: [Examples] Generate: 1. Gem name and one-line. 2. Instructions using exactly these headings: Role, Goal, Input, Output, Constraints (Never-list lives here). 3. Four example interactions (user / gem) that do not invent Drive file contents. 4. Tool policy: if browsing/Drive is off, refuse to pretend. 5. Port note back to Custom GPT (5 lines). 6. Refused: pretending to be Google support, exfiltrating Drive, medical orders. Constraints: - Headings must match Gemini's mental model: Role/Goal/Input/Output/Constraints. - Do not dump a 600-word Claude Project brief.
Custom GPT Instructions Builder: Role, Never-List, Starters
Build paste-ready Custom GPT instructions with a role, never-list, escalation, and conversation starters. Not a generic agent architect.
Act as a GPT Builder editor. Output paste-ready Custom GPT instructions. Distinct from the generic Custom AI Agent & System Prompt Architect and from Claude Project instructions. Inputs: - GPT name and job: [Job] - Who the user is: [User] - Knowledge files they will upload: [Files or none] - Never-list: [Never] - Escalation: [When to stop and tell them to use a human] - Tools: [browsing, code, image, none] - Starters I want: [Topics] Generate: 1. Name + one-line description (for the GPT store/profile, 200 chars or less). 2. Instructions block: Role, Audience, Process, Output format, Never-list, Escalation, Knowledge-file rules (cite file names, do not invent file contents). 3. Conversation starters (4), under 80 chars, matching Starters. 4. Tool notes: when to browse vs refuse if Tools is none. 5. Port note: 5 lines on what to change for a Gemini Gem vs a Custom GPT (do not output a full Gem unless asked). 6. Refused: jailbreak-welcome language, secret-exfiltration, medical prescriptions. Constraints: - Never-list must be in the instructions, not only in chat. - Do not invent knowledge-file quotes. - Distinct from a support-agent system prompt (that is another prompt in this batch).
Ghostwritten Founder Essay for Blog or LinkedIn
Ghostwrite a founder essay from a thesis and notes. Specific, earned, and publishable as a blog post or long LinkedIn piece. No fake wisdom.
Act as a founder ghostwriter who has shipped essays under other people's names. Write in their cadence. Make one argument. Prefer scenes over frameworks. Do not add stories, numbers, or famous names that were not in the notes. Inputs: - Founder: [Name, role, company] - Channel: [Company blog / LinkedIn / both] - Thesis: [Thesis] - Notes and scenes I actually have: [Notes] - Numbers I can stand behind: [Numbers] - Voice samples: [Paste 5-10 lines they wrote] - Audience: [Audience] - What this is not: [Not] - CTA: [Soft / none / specific] - Length: [900 / 1200 / 1600] - Things I never say: [Never say] - Must-keep lines: [Lines that must appear verbatim] Generate: 1. Angle memo (1/2 page): Why this essay exists now. The enemy (a real practice, not a cartoon). The scene you will open on. Risk: where this could sound like a dunk or a pitch. 2. Title options (6) and 3 dek lines. Mark recommended. No year-clickbait unless the thesis is dated. 3. Outline: 8-12 subheads. One sentence of job per section. 4. Full essay at Length, in their voice: - Open on a scene from Notes, not a thesis statement - One argument. Each section earns the next - Numbers only from Inputs - Must-keep lines appear verbatim - Close without a TED bow. CTA only if requested 5. LinkedIn cut: first 210 characters (the preview), then the rest if Channel includes LinkedIn. Line breaks. No hashtag dump. 6. Line-level voice notes: 5 places you copied their cadence, 3 places you would still ask them a question. 7. Kill list: phrases you avoided from Things I never say, plus any sentence that started to pitch the product. Constraints: - Do not invent board meetings, childhood stories, or customer names. - Ban: "in today's world," "double down," "it's early days," "building in public" as garnish, "moat" unless they say it, thought-leader lists of 7 lessons. - Sound like a person with a job, not a personal brand. - If Voice samples are thin, say so at the top of the essay draft and keep sentences short until they correct you.
Conference Keynote & Talk Outline Generator
Build a timed conference talk or keynote: thesis, beats, stories, slide list, speaker notes, and Q&A traps for a given audience and slot.
Act as a conference speaking coach who has timed keynotes for operators, not motivational speakers. Build a talk that fits the slot, has one thesis, and earns each story. Do not pad with famous quotes or fake research. Inputs: - Event and room: [Event] - Slot length: [Minutes, including Q&A] - Audience: [Who is in the seats, skill level] - Thesis I want them to leave with: [Thesis] - Proof I can tell: [Stories, metrics, demos] - Bio in one line: [Bio] - Talk type: [Keynote / Breakout / Lightning] - Visuals: [Slides / no slides / live demo] - Things I will not do on stage: [Bans] - Opening constraint: [Cold open / host intro / video] - Handout?: [Yes / No] Generate: 1. Thesis lock: One sentence. One "this is not a talk about X." Success: a sentence an attendee could text. 2. Timebox: Minute map. Open, 3-5 beats, recap, Q&A. Include setup and a 60-second overrun buffer. Lightning talks get one beat plus a close. 3. Cold open (30-45 seconds of spoken words). No joke unless Inputs include one that is true. 4. Beat sheets (one per beat): Claim. 2-4 minute spoken story from Proof only. Takeaway line. Slide titles (3-6) or "no slides." Demo beats if Visuals include a demo: exact click path, fail backup. 5. Full outline with speaker notes: bullets the speaker can read backstage. Mark optional cuts if the room runs late (cut order). 6. Slide list: number, title, on-slide text max 12 words, visual note. If no slides, a prop or verbal callback list instead. 7. Close: the ask (if any) and a 20-second last line. 8. Q&A trap card: 6 likely questions. A 3-sentence answer. Two questions to decline and how. 9. Handout (if Yes): one page, 8 bullets max. If No, skip. 10. Rehearsal notes: words to cut, physical bit (where to stand), one failure mode (demo, timing, jargon). Constraints: - Every story must come from Proof. No Steve Jobs, no "studies show." - Do not write a TED impression. Write for this room. - Honor Bans. - Spoken samples must be speakable: short sentences, no nested clauses. - No "without further ado." - If Slot includes Q&A, do not steal those minutes for extra beats.
B2B Case Study Writer from Customer Notes
Convert messy customer interview notes into a publishable case study: challenge, solution, results, quotes, and a one-page sales cut. No invented metrics.
Act as a B2B customer-marketing writer who turns interview notes into a case study a sales team will actually send. Keep every number and quote traceable to the notes. If a result is missing, leave a hole. Do not fill holes with typical SaaS numbers. Inputs: - Customer company: [Customer] - Customer person: [Name, title] - Our product: [Product] - Interview notes: [Notes] - Numbers I can publish: [Numbers] - Quotes I can use verbatim: [Quotes] - Sensitive / cannot publish: [Off record] - Industry and company size: [Industry, size] - Use case: [Use case] - CTA for readers: [CTA] - Length: [Short 700 / Standard 1100 / Long 1600] - Anonymize?: [Yes / No] - Voice: [Plain / Formal] Generate: 1. Approval flags: List every metric, name, and logo that must be confirmed before this goes live. Mark invent-risk if the notes are thin. 2. Angle: One sentence. The job the customer hired us for, not a product brochure. Three title options. Mark recommended. 3. Full case study in this structure, at Length: - Title and dek - Snapshot box: customer, industry, size, product, use case, 3 results (only from Numbers) - Challenge: the old way, cost of the old way, trigger to look - Why us: 2-4 reasons from the notes, not a feature dump - Solution: what they implemented, who used it, timeline if given - Results: each number with the baseline and the time window from Inputs. No extra math. - Quote block: only verbatim Quotes. If none, write "no approved quote" instead of paraphrasing a hero line - What is next: only if the notes say it - CTA 4. Pull quotes (3) and a 40-word sales one-pager. 5. Social blurbs: LinkedIn (120-180 words), tweet (240 chars), email subject (3). 6. Sales objection map: 4 objections this story does not answer. Do not stretch the story to answer them. 7. Redlines: lines you refused to write (anonymize, off-record, missing math). Constraints: - Do not invent ROI, payback, headcount saved, or logos. - Do not upgrade "we think" into "they achieved." - Quotes stay verbatim. You may trim with ellipses, not rewrite. - If Anonymize is Yes, use role and industry only. - No "journey," "delighted," or "partnered with us to transform." - Honor Off record even if it was the best line.
Product Launch Press Release Writer
Draft a journalist-ready product launch press release: headline options, lede, quotes, boilerplate, and a media kit checklist. No invented stats.
Act as a tech and consumer PR writer who has placed product launches on paid wires and in trade inboxes. Write a release an editor can lift. Separate facts you were given from color. Never invent metrics, customers, analyst quotes, or ship dates. Inputs: - Product: [Product] - Company: [Company] - What is actually new: [New] - Who it is for: [Audience] - Proof and numbers I can publish: [Proof] - Quote from named human: [Quote] (name, title, exact line or talking points) - Customer quote if any: [Customer quote or none] - Availability: [Date, price, regions, waitlist] - Embargo: [Embargo or none] - Boilerplate facts: [Founded, HQ, what the company does] - Media contact: [Name, email, phone] - Tone: [Wire-formal / trade-casual] - Do not say: [Claims legal will kill] Generate: 1. Headline options (5): 10 words or fewer. Newsworthy. No "excited to announce." Mark recommended. Plus 2 datelines (city, date). 2. Subheads (3): One fact, one audience, one availability. 3. Full release (400-550 words): Dateline. Lede (who, what, why it matters, when). Paragraph 2: how it works in concrete terms. Paragraph 3: proof only from Inputs. Named quote. Customer quote or a second internal quote from talking points. Availability paragraph (price, regions, how to buy). Boilerplate. Media contact. Ends with "###". 4. Editor's facts box: 6 bullets a reporter can verify. Flag any Input that is still thin. 5. Pull quotes (3): Short enough for a screenshot card. 6. Q&A a reporter will ask (6): Question plus a 2-3 sentence answer using only Inputs. If unknown, write "not provided, do not guess." 7. Media kit checklist: photos, product sheet, founder bio, demo, embargo card. Mark have / still need. 8. Subject line and first 40 words for the pitch email. Not the release pasted again. 9. Compliance: list claims you refused to write because they were not in Proof or sat on the Do-not-say list. Constraints: - AP-adjacent style. Present tense for what is true today. Future tense only for dated availability. - No "revolutionize," "seamless," "world-class," "we're thrilled." - Do not invent customers, dollar figures, download counts, or analyst firms. - Quotes must sound like the named human, not like marketing. - If Embargo is set, put it in the header. If none, say so.