Agent Requirements

Agent Objective:

So what the agents will do is porcess the context that is recived from the vectore database plus the question the users asked.

ones the answer is recived a hullucination detection agant will check if the answer is correct and relevent with the document that is recived from the vector database and aplpy cititions if needed

Autonomy & Constraints:

predefined context will only allow it to answer form the context that is provided

Tool & Action Design

at the moment the llm2 will send back a json object with the retry function specifded if it evalutes that the response from the llm which generated the answer is not consists with the pre defined rules.

Agent Architecture

so the ui send a request with the ui question the api backend calls the vector database using semtaic + norma search retrive the relevent documents with there citation infromaation sends it to the llm to genearte an answer another one which is llm2 will check if it meets the predefined rules and strture and if yes procceds to send back the response if not then sends a response back to api which has a function called retry or a functoîon call termiate if the question or the anser vilates anything

Memory & Context Design

How does the agent manage memory and context? Define short-term (conversation) and long-term (persistent) memory strategies. How is context window managed? What retrieval mechanisms are used (RAG, vector stores, summarization)?

Core Reasoning

Deep dive into the agent's reasoning strategy. How does it plan, decide, and self-correct? Discuss prompt engineering, chain-of-thought, reflection loops, failure handling, and tradeoffs.

Safety & Permissions

What guardrails and safety mechanisms does the agent have? Define permission boundaries, human-in-the-loop checkpoints, content filtering, rate limiting, and how the agent handles adversarial inputs or unexpected states.