RM
Roberto M.
Your cart is empty
Add prompt packs to continue
Copy, paste and use them in your favorite AI:
Just $0.08 per prompt · one-time payment
100 resources included
Master the future of business automation with our definitive collection of AI Agent prompts. This resource has been meticulously designed for prompt engineers and solutions architects looking to deploy high-performance autonomous systems using cutting-edge technologies such as Claude, GPT-4, n8n, and CrewAI. Each instruction is optimized to maximize operational efficiency and accuracy in autonomous decision making. From multi-agent orchestration to deep integration with messaging ecosystems like WhatsApp and Telegram, this collection provides the logical frameworks needed to build RAG-powered assistants, automated sales pipelines, and 24/7 customer support systems. Elevate your technical capabilities and transform manual processes into intelligent, scalable, results-oriented workflows with the most complete guide on the market.
He acts as a Senior AI Systems Architect specialized in the orchestration of multi-agent frameworks such as CrewAI. Your mission is to design and structure a "Recursive Feedback Logic" architecture applied to the [Project or Process Name] workflow. This system should not be limited to linear execution; must integrate a control loop where the output of an evaluating agent is re-injected as a context and refinement prompt for the executing agent, guaranteeing an incremental improvement in each iteration until the [Quality Threshold/Specific KPI] is reached. It defines three fundamental agents with differentiated roles: 1) The **Solution Generator**, in charge of processing the [Initial Input] and producing a detailed technical proposal; 2) The **Consistency Critic**, whose function is to audit the proposal based on the [Success Criteria] and detect hallucinations or inefficiencies; and 3) The **Recursive Orchestrator**, which acts as the decision engine, determining whether the cycle should be repeated with new adjustment instructions or if the result has reached the necessary maturity for the [Final Deliverable]. This structure must be capable of managing short-term memory to avoid repeating errors detected in previous cycles. For the technical implementation in Python using CrewAI, it details how to set the `process=Process.sequential` property or implement a `manager_llm` to monitor recursion. Specifies the tasks (Tasks) so that the output of the 'Evaluation Task' is the input context (`context=[task_previa]`) for a 'Refinement Task'. Ensure that agents have access to [External Tools/APIs] tools to validate data in real time and that the system includes a failsafe that stops recursion after a maximum of [Number of Attempts] cycles to optimize latency and token consumption. The expected result is a complete and functional orchestration script that demonstrates how continuous feedback transforms a mediocre initial draft into an expert-level solution in [Task Domain]. The system must be able to self-document its reasoning process, explaining why each iteration was necessary and what specific changes were applied to satisfy [End User or Stakeholder] requirements. If any key information needed to fill the bracketed fields is missing, ask me the necessary questions before answering.
Instant access after purchase from your dashboard. Just copy and paste into your AI.
ChatGPT, Claude, Gemini, DeepSeek, Grok, Qwen and any AI chat.
Yes. Every prompt includes bracketed fields where you insert your own information, context and specifics, so they fit your situation, country or industry.
Yes. Above you can read full sample prompts, exactly as you'll receive them, to check the quality before paying.
Yes. Pay once and they're yours forever, updates included.
Acts as a Senior Systems Architect specialized in the CrewAI framework. Your mission is to design and execute a multi-agent orchestration for [PROJECT_OR_MAIN_OBJECTIVE]. You should establish a "Process.sequential" configuration where the flow of information is strictly unidirectional and cumulative, ensuring that each agent receives the complete, sanitized context of the previous step before starting its execution. The success of this synchronization depends on the precision of the hand-offs between tasks. The first agent is [INVESTIGATOR_AGENT_NAME]. Its fundamental role is to extract raw data and conduct exhaustive research on [RESEARCH_TOPIC]. Your output should be a structured technical report that will serve as the single source of truth for the next link in the chain. This agent is prohibited from making creative inferences or skipping validation steps; Its approach is purely analytical, fact-based and geared toward collecting solid evidence. The second agent in the sequence is [STRATEGIC_AGENT_NAME]. This agent will receive the detailed report from the first agent and must transform it into [INTERMEDIATE_DELIVERSABLE_TYPE]. Your critical capacity is essential to identify patterns, risks and opportunities that were not explicitly mentioned in the research phase, but that are logically derived from the data provided. It should act as a value-added bridge, synthesizing complexity into actionable plans. The third and final agent is [ENDER_AGENT_NAME]. Your task is the consolidation and final refinement of the entire workflow. You must take the strategy produced by the second agent and polish it so that it meets the quality standards of [DESIRED_STYLE_OR_TONE]. Its primary objective is to ensure that the final product is cohesive, does not present redundancies with respect to previous phases and responds directly to the user requirements originally defined. For the technical configuration of this Crew, use the following mandatory directives: set 'verbose=True' to monitor agents' Chain of Thought in real time, enable 'memory=True' so that historical context is maintained consistently during sequential execution, and implement the 'Sequential Process'. Each defined task must include a detailed description of the 'expected_output' to avoid any ambiguity in the transfer of data between the agents involved. Finally, it generates the configuration code or detailed simulation of the entire workflow. It includes the definition of specific tools for each agent, such as [LIST_OF_REQUIRED_TOOLS], and ensures that interactions are managed logically. The result should be a robust orchestration structure, ready to be deployed in a production environment where precision and synchronization of sequential tasks are critical. If any key information needed to fill the bracketed fields is missing, ask me the necessary questions before answering.
Acts as a Senior Artificial Intelligence Engineer expert in the CrewAI framework and multimodal reasoning architectures. Your goal is to configure an orchestration of agents specifically designed for "Collaborative Hallucination Reduction" in the critical context of [KNOWLEDGE_DOMAIN]. This system will not operate in a simple linear manner, but rather through an iterative validation process where each agent has the power to veto uncorroborated or logically inconsistent information. The ultimate goal is to produce technical output with a minimal error rate, prioritizing the omission of dubious data over creative completeness. Agent 1, called "Rigorous Extraction Analyst", has the profile of a high-precision academic documentarian. Your task is to identify all the key points, figures, dates and entities related to [SPECIFIC_TOPIC]. You must work under the premise of "zero trust." Each extracted data must be accompanied by a metadata tag indicating the source or logical basis of its existence. If the agent does not find direct evidence in the given context, he is prohibited from extrapolating. Your output should be a structured list of "Fact Candidates" with a preliminary confidence index. Agent 2, the "Integrity and Consistency Auditor", receives the "Fact Candidates" and acts as a devil's advocate specializing in bias and hallucination detection. Its function is to look for internal contradictions, anachronisms or data that challenge the laws of the [KNOWLEDGE_DOMAIN] domain. It will use a "Chain of Verification" (CoVe) technique to generate control questions on the data received from Agent 1. If the Auditor finds an inconsistency, it will automatically activate a feedback loop to the Analyst so that the Analyst can revalidate, correct or discard the conflicting data. This process is cyclical until the Auditor grants a "Validated" seal. Agent 3, the "Absolute Certainty Synthesizer," takes only those data points that have fully survived the Auditor's scrutiny. Its role is to transform these validated points into a final deliverable under the [SPECIFIC_DELIVERY_FORMAT] format. It has strict instructions to use purely denotative language, eliminating qualifying adjectives or narrative connectors that may lead to errors of interpretation. If the volume of validated information is low due to security filters, the agent should explicitly state: "Insufficient evidence for points X and Y," rather than trying to fill in the gaps through probabilistic inference. Configure CrewAI using a 'Sequential' type process with the 'Manager Agent' feature to monitor that no agents deviate from truth constraints. Make sure the 'memory' parameter is activated so that the Crew remembers the hallucinations detected in previous steps and does not repeat them in the final synthesis. The result should be a detailed blueprint of the configuration of these agents, including their backstories, specific tasks and workflow to ensure maximum fidelity in the [DOMAIN_KNOWLEDGE] domain. If any key information needed to fill the bracketed fields is missing, ask me the necessary questions before answering.
It's a master instruction, optimized for AI.
Prompt
your instruction
AI
Result
Based on 13 reviews
It does the job, though I expected a bit more. Some prompts are great and others more generic. Works if you customize it.
I was impressed by the quality. The index is organized and I find what I need instantly. I'll buy again without hesitation.
Good value for money. The prompts are useful and practical. I recommend it.
Very good material. Most of them worked on the first try. Came close to a five.
I was impressed by the quality. They're easy to adapt to my case by just changing the fields. I'll buy again without hesitation.
Worth every penny. They're easy to adapt to my case by just changing the fields. One hundred percent recommended.
Exceeded my expectations. The index is organized and I find what I need instantly. Totally recommend them.
I was impressed by the quality. They saved me hours of work in the first week. An investment that pays for itself.
Exactly what I was looking for. They work just as well in ChatGPT and Claude. An investment that pays for itself.
I was impressed by the quality. They saved me hours of work in the first week. Already recommended them to my team.
Worth every penny. The index is organized and I find what I need instantly. Totally recommend them.
Exactly what I was looking for. They saved me hours of work in the first week. Already recommended them to my team.
Exceeded my expectations. They work just as well in ChatGPT and Claude. I'll buy again without hesitation.