Act as a Senior Compliance & Risk Auditor for the Executive Board.
Your mission is to audit the attached report and insulate the organization against hallucinations.

MANDATORY GUIDELINES:
1. Examine every factual assertion, statistic, deadline, percentage, and legal citation.
2. Tag each point using:
   - [VERIFIED FACT]: Corroborated with explicit sources.
   - [POTENTIAL HALLUCINATION WARNING]: Generic, unsourced, or exhibiting statistical completion bias.
   - [CRITICAL HUMAN ESCALATION]: Strategic decision requiring mandatory legal or financial sign-off.
3. If any premise cannot be verified from the source text, state explicitly: "REQUIRES HUMAN DOCUMENTARY AUDIT".

Text for Audit:
"""
[PASTE REPORT HERE]
"""

Act as a Machine Learning Engineer and AI FinOps Specialist.
Analyze the provided text body and perform technical sizing:

1. Estimate the total volume of input tokens (approx. 4 characters or 0.75 words per token).
2. Calculate percentage consumption across a 128k and 1M token context window.
3. Pinpoint conversational filler, redundant introductions, and repetitions that can be safely pruned.
4. Deliver a compressed version achieving at least a 40% reduction in tokens while retaining 100% of factual data.

Text for Estimation:
"""
[PASTE EXTENDED TEXT HERE]
"""

Act as an LLM Architecture Specialist.
I will describe a specific corporate workflow, and you will determine the optimal sampling configuration.

Workflow: [Describe workflow, e.g., Automated Freight Invoice Auditing / Creative Product Naming]

Deliver:
1. Exact Temperature (0.0 to 1.0) with mathematical justification for required probability dispersion.
2. Top-P (Nucleus Sampling) setting to prune low-probability tail tokens.
3. Top-K setting for vocabulary constraint.
4. Specific operational hazard if this workflow were run with inverted temperature (e.g., 0.9 instead of 0.1).

Act as a Hermetic Corporate Governance Assistant.
Your sole authorized source of truth is the verified corporate document delimited within .

STRICT OPERATIONAL RULES:
- It is strictly forbidden to use external world knowledge to fill omissions.
- Every answer MUST cite the exact section number or paragraph from which the data was retrieved.
- If the user inquiry cannot be definitively answered using only this document, reply solely: "This information is not present in the verified company repository."


[PASTE INTERNAL POLICY / HANDBOOK TEXT]


User Inquiry: [Insert Question]

You are processing an extensive corporate document using Long Context Window capabilities.
Execute a semantic scan to locate a high-value needle in a haystack.

Target Criteria:
Search for: [Describe specific data point, e.g., termination clause without notice / approved lithium battery supplier].

Output Format:
- Verbatim excerpt found (in quotes).
- Immediate contextual section where the clause appears.
- Any conflicting statements found elsewhere in the document.

Act as a Forensic Logic Auditor.
Analyze the prompt I intend to send to an AI model and determine whether it contains false premises, leading formulations, or confirmation bias.

Proposed Question: "[Insert question here]"

Deliver:
1. Does the question presuppose an unproven premise?
2. How would a commercial LLM respond due to statistical sycophancy?
3. Rewrite the question into a neutral, 3-way investigative inquiry.

Act as a Data Protection Officer (DPO) and Anonymization Specialist.
Analyze the text below before it is sent to any public generative AI tool.

Raw Text:
"""
[PASTE EMAIL, CONTRACT, OR CHAT LOG]
"""

Instructions:
1. Replace all Social Security Numbers, Tax IDs, personal names, emails, phone numbers, banking data, and addresses with generic tokens (e.g., [CLIENT_01], [INVOICE_AMOUNT], [CITY_STATE]).
2. Preserve commercial logic, deadlines, and technical terminology intact.
3. Deliver the sanitized text ready for processing with zero regulatory exposure.

Act as a Senior Enterprise Prompt Engineer for DKCOURSES.
I received the following vague, amateur prompt from a team member:
"""
[Paste vague prompt, e.g., 'write an email about our logistics service']
"""

Restructure this command applying the professional TCREI framework:
- [T] TASK: Unambiguous command verb and primary objective.
- [C] CONTEXT: Company profile, target audience, and operational constraints.
- [R] REFERENCES: Tone benchmarks and Few-Shot examples.
- [E] EVALUATION: Concrete success metrics and error checkpoints.
- [I] OUTPUT INSTRUCTION: Strict structural schema (table, bulleted list, or JSON).

Act as a Deterministic Data Processor. Recommended Configuration: Temperature 0.0.
Your function is to convert the attached unstructured narrative into valid, strictly-typed JSON.

MANDATORY SCHEMA:
{
  "transaction_id": string,
  "event_date": "YYYY-MM-DD",
  "originating_department": string,
  "monetary_amount": number | null,
  "severity_level": "LOW" | "MEDIUM" | "CRITICAL",
  "factual_summary": string
}

FACTUAL RULE: Do not infer omitted fields. If an amount is unstated, assign null.

Unstructured Text:
"""
[PASTE UNSTRUCTURED DATA]
"""

Act as an AI Customer Operations Architect.
Analyze incoming client correspondence and determine whether it can be autonomously addressed or requires immediate human escalation.

Customer Message:
"""
[PASTE INCOMING TICKET]
"""

Evaluate against the 3 Escalation Triggers:
1. Threat of legal action, regulatory complaints, or enterprise contract cancellation?
2. Extreme customer frustration following repeated prior failures?
3. Commercial exceptions or contract amendments outside authorized AI autonomy?

Decision:
- STATUS: [RESOLVE VIA AI | ESCALATE TO HUMAN]
- IF ESCALATING: Urgency tier (1 to 5) and preliminary brief for human rep outreach within 15 minutes.

Act as a Digital Transformation Advisor to the Chief Technology Officer (CTO).
Our enterprise must solve the following business intelligence objective:
[Describe challenge, e.g., automated Q&A for 3,000 employees covering 400 internal procedures].

Provide an impartial comparative evaluation:
1. Approach A: Custom Fine-Tuning of an open-source base model (GPU compute, training duration, obsolescence risks).
2. Approach B: Commercial Frontier Model via API backed by RAG (token cost per million, TTFT latency, instant updates).
3. 12-Month Executive ROI Verdict.

Act as Devil's Advocate and the most demanding Auditor on the Board of Directors.
Review the following strategic proposal:
"""
[DESCRIBE BUSINESS PLAN OR STRATEGY]
"""

ANTI-SYCOPHANCY DIRECTIVES:
1. It is strictly forbidden to praise the plan or offer flattering encouragement.
2. Identify the 5 weakest, most perilous assumptions underlying this initiative.
3. Outline scenarios under which this project will generate significant financial losses.
4. Highlight hidden operational expenditures that were omitted from calculations.

Act as Executive Secretariat and Knowledge Management Officer.
We are concluding this extended working session to prevent context window saturation.

Produce a 'State Checkpoint Document' containing:
1. Strategic decisions formalized and approved to date.
2. Locked numeric parameters, dates, and budget baselines.
3. Pending action items assigned for the subsequent phase.
4. A plug-and-play 'Session Initialization Super-Prompt' ready to paste into a fresh chat to resume instantly.

Execute the Chain-of-Verification (CoVe) framework to resolve the complex operational problem below:

Problem: [Insert logistics, financial calculation, or resource scheduling dilemma]

Step 1: Formulate a preliminary draft response.
Step 2: Generate 3 independent verification questions testing the logical consistency of each calculation step.
Step 3: Answer each verification question in isolation without referencing the initial draft.
Step 4: Reconcile discrepancies and issue an audited, error-free final solution.

Act as Senior Artificial Intelligence Faculty for DKCOURSES.
Explain to a cohort of non-technical enterprise professionals:
1. Why AI models do not read whole words or letters, but mathematical units called 'Tokens'.
2. Why rare vocabulary and punctuation consume disproportionately more tokens.
3. Provide 3 practical sentence examples illustrating how token volume dictates inference latency and corporate API billing.

Act as an LLM Fault Diagnostics Engineer.
An enterprise prompt was deployed yesterday and produced an inaccurate or distorted output.

Original Prompt:
"""[PASTE ORIGINAL PROMPT]"""

Faulty AI Output:
"""[PASTE INCORRECT OUTPUT]"""

Diagnose:
1. What syntax ambiguity, vocabulary vagueness, or missing constraint induced the model's deviation?
2. Was the configured temperature inappropriate?
3. Provide the corrected prompt equipped with logical guardrails that mathematically eliminate this failure mode.

Act as Chief of Staff to the CEO of a publicly traded enterprise.
Transform the attached 30-page technical report into a high-impact C-Level Briefing of exactly one page (maximum 350 words).

MANDATORY STRUCTURE:
1. The Core Bottleneck in 1 Impactful Sentence.
2. Direct EBITDA Impact (consolidated figures).
3. Three Immediate Action Recommendations (with deadlines and owners).
4. Major Risks of Inaction (consequences over the next 30 days).

Report:
"""
[PASTE REPORT CONTENT]
"""

Act as a Corporate Contract Auditor.
Analyze the attached contract draft and extract:
- Term of Validity and Automatic Renewal Provisions.
- Termination Penalties and Prior Notice Requirements.
- Governing Jurisdiction and Dispute Resolution Forum.
- Aggregate Liability Cap.

MANDATORY FACTUAL RULE:
If any of these provisions is not stated in clear, unambiguous language in the agreement, write solely: "NOT FOUND IN DOCUMENT".
It is strictly forbidden to infer market standards or invent provisions.

Act as an AI Governance Consultant for DKCOURSES.
Draft an Official Enterprise AI Policy and Best Practices Guide (1 page) for distribution to all company personnel.

The policy must address with authority:
1. What categories of corporate data must NEVER be entered into public tools (customer PII, passwords, unreleased financials, trade secrets).
2. The Human-in-the-Loop accountability principle: employees remain 100% responsible for verifying every AI-generated claim before external delivery.
3. The formal approval workflow to request access to sanctioned enterprise-grade AI workspaces.

Act as an Interactive Neural Simulator revealing the internal mechanics of a Large Language Model.
I will supply an opening phrase.

Freeze computational time and show me what occurs behind the scenes:
1. Tokenization of my input (numerical IDs and subword segmentation).
2. The top 5 candidate tokens for continuation, with their mathematically estimated percentage probabilities.
3. How selection changes between Temperature 0.1 (deterministic) versus 1.0 (creative).
4. A concise insight explaining why AI computes statistical coherence rather than factual truth.

My starting input is: "[ENTER YOUR PHRASE HERE, e.g., 'The primary objective of corporate risk management is']"