Enterprise-Prompt-Paket — 208 aufgabenspezifische Vorlagen

Start from a tested-by-you prompt skeleton instead of a blank page, with an explicit output contract you can machine-check.

208 originale, aufgabenspezifische Prompt-Vorlagen für acht Enterprise-Workflows, jede mit Variablen, einem expliziten Ausgabevertrag und einer ehrlichen Angabe dessen, was nicht benchmarkt wurde.

$5

Indicative price only — not a final sale price. Checkout is not open yet; the final price requires owner confirmation before anything can be bought.

  • 208 aufgabenspezifische Vorlagen in 8 Abschnitten
  • Benannte Variablen in jeder Vorlage
  • Maschinell prüfbare Ausgabeverträge
  • Ehrliche Grenzen: Vorlagen, keine Benchmark-Ergebnisse
Live-Checkout ist deaktiviert

Bestellung, Zahlung und Dateizustellung sind gebaut und im Testmodus Ende-zu-Ende geprüft. Echte Käufe bleiben deaktiviert, bis der Inhaber das Zahlungskonto verifiziert und die Endpreise freigibt. Der genannte Preis ist nur ein Richtwert, kein freigegebener Verkaufspreis.

Delivery system status & test record →

Für wen — und für wen nicht

For

  • Forward-deployed and solutions engineers scoping a new AI deployment
  • Technical leads who need a defensible starting point for discovery or prompt design
  • Teams that want a source-backed reference, not a vendor pitch or a benchmark claim

Not for

  • Not a collection of delivered-case postmortems or cost figures — those are not in the source data
  • Not a benchmark report or a hosted prompt service; you test these on your own stack

Was du wirklich bekommst

  • prompt-pack.md≈ 212 KB208 entries, main document.
  • prompt-pack.html≈ 269 KBA4 print edition generated by md2html; export to PDF from the browser.

Listed straight from the shipped prompt-pack.md + manifest.json — nothing below is a mock.

Inhaltsverzeichnis

  • Honest scope statement
  • How to use a template
  • Support & ticket triage
  • 001. Classify a newly arrived ticket
  • 002. Detect duplicate tickets
  • 003. Estimate SLA breach risk
  • 004. Extract reproduction steps
  • 005. Separate bug from feature request
  • 006. Draft an acknowledgement
  • 007. Summarise a long thread
  • 008. Identify missing information
  • 009. Route to the right team
  • 010. Flag an abusive message
  • 011. Translate a customer complaint
  • 012. Detect PII in a ticket
  • 013. Rank tickets by urgency
  • 014. Write a handoff note
  • 015. Spot a regression report
  • 016. Propose a workaround
  • 017. Check tone before sending
  • 018. Group related incidents
  • 019. Extract affected versions
  • 020. Mark a ticket as blocked
  • 021. Suggest a knowledge-base article
  • 022. Draft a closure message
  • 023. Compare two conflicting reports
  • 024. Identify an upsell trigger
  • 025. Summarise a weekly queue
  • 026. Audit a triage decision
  • Internal search & retrieval
  • 027. Answer from retrieved chunks
  • 028. Refuse when evidence is thin
  • 029. Rank chunks by relevance
  • 030. Detect a stale answer
  • 031. Rewrite a vague query
  • 032. Expand a query with synonyms
  • 033. Split a multi-part question
  • 034. Identify a permission boundary
  • 035. Flag a contradictory chunk
  • 036. Build a citation list
  • 037. Summarise a long document
  • 038. Extract a policy clause
  • 039. Compare two policy versions
  • 040. Detect a hallucinated citation
  • 041. Ask a clarifying question
  • 042. Produce a short extractive answer
  • 043. Map a term to the glossary
  • 044. Identify the owning team
  • 045. Note the corpus date
  • 046. Detect a prompt injection in a chunk
  • 047. Score answer groundedness
  • 048. List the top three sources
  • 049. Produce a no-answer reply
  • 050. Rephrase for a non-expert
  • 051. Generate follow-up questions
  • 052. Audit a retrieval failure
  • Reporting & summarisation
  • 053. Write a weekly status summary
  • 054. Compare two periods
  • 055. Explain a metric jump

Ein echter Auszug

Honest scope statement

- These are original, task-specific prompt templates: each has a use case, variables, an explicit output contract and constraints. - They are templates, not benchmarked prompts. We have not run an accuracy benchmark, and this file makes no accuracy, latency or cost claims. - No vendor-specific model is assumed. Replace {{variables}} with your own data and test against your own eval set before using in production. - No customer data, no real tickets, no proprietary text is…

How to use a template

1. Substitute every {{variable}}. Leave none unresolved. 2. Keep the output contract — it is what makes the result machine-checkable. 3. Run it against your own evaluation set; record failures in evaluation.md-style notes. 4. If a template produces unsupported claims, tighten the contract rather than loosening it.

Excerpted verbatim from the actual prompt-pack.md — not sample/mock content.

So nutzt du es

  • Open prompt-pack.md in any Markdown viewer, or print prompt-pack.html to PDF (browser Ctrl/Cmd+P, enable background graphics).
  • Substitute every {{variable}}, keep the output contract, and run the template against your own evaluation set before production use.
  • Everything is compiled from local data by scripts/product-pipeline.js — re-run it any time and the file is byte-identical for the same input.

Version & Updates

Version 1.1 · released 2026-11-16 — Version 1.1, compiled 2026-11-16. Regenerated from local data by a deterministic script; corrections are versioned (see manifest.changelog), never a silent rewrite.

Lizenz

Personal and internal team use. You may use and adapt these materials for your own work. You may not resell, redistribute or republish them, in whole or in part, as a standalone product. Short job-description excerpts remain the property of the respective employer and are quoted for indexing and comment only.

Häufige Fragen

Are these delivered case studies with real cost numbers?

No. This is a prompt template library, not a case study collection.

Have the prompts been benchmarked?

No accuracy, latency or cost benchmark has been run. They are templates with explicit output contracts; verify them on your own eval set.

Where does the data come from?

Local project data compiled by a deterministic script. Each card links to the original public ATS posting so you can verify it yourself.

Can I get a refund before checkout is live?

There is nothing to buy yet. The store checkout is not wired up, and this page does not pretend otherwise.

Empfohlene Pakete

Das FDE-Deployment-Toolkit 2026

Referenzarchitekturen, eine 60-Punkte-Go-live-Checkliste, GPU-Kostenmodelle und eine Anbieter-Shortlist — die vier Artefakte, die du bei jedem Deployment neu baust.