Recherche Entreprise Mcp

What is Recherche Entreprise Mcp

recherche-entreprise-mcp is a project designed to enhance job searching by automating processes using MCP, Astra, and Arize. It provides pre-written prompts to help users practice coding from home, specifically focusing on the integration of various APIs and data storage solutions.

Use cases

Use cases include automating job searches for specific companies, analyzing company data with filters, storing search results in a NoSQL database, and monitoring the performance of AI models in real-time.

How to use

To use recherche-entreprise-mcp, you can create a Symfony command called ‘app:recherche-entreprise’ that utilizes the Recherche d’Entreprises API. You can execute commands through the console to perform basic searches, apply filters, and manage pagination for results.

Key features

Key features include the ability to perform basic and filtered searches for companies, integration with Astra for NoSQL data storage, and performance monitoring of AI models using Arize. The project also supports various command options to customize search results.

Where to use

recherche-entreprise-mcp can be used in fields such as job searching, data analysis, and AI model performance monitoring. It is particularly useful for developers and data scientists looking to automate and enhance their workflows.

Voici des prompts pré-écrit qui vous permettrons de vous entrainer chez vous dans votre canapé pour expérimenter le Vibe Coding.

Créer un serveur MCP en PHP

Je veux créer un serveur MCP avec PHP et Symfony (utiliser HTTP Client) en suivant le protocole décrit dans cet article [MCP : Le protocole open-source qui transforme les chatbots LLM en agents intelligents](https://jolicode.com/blog/mcp-the-open-protocol-that-turns-llm-chatbots-into-intelligent-agents).

Intégrer une commande Recherche Entreprises

Créer une commande symfony app:recherche-entreprise utilise l'api Recherche d’Entreprises grâce à la [structure Open API mise à disposition](https://recherche-entreprises.api.gouv.fr/openapi.json).

Exemple de commandes à prompter :

• Recherche basique : php bin/console app:recherche-entreprise "La Poste"
• Recherche avec filtres : php bin/console app:recherche-entreprise "La Poste" --categorie-entreprise=GE --region=11
• Recherche avec réponse minimale et champs spécifiques : php bin/console app:recherche-entreprise "La Poste" --minimal --include=siege,complements
• Recherche avec pagination : php bin/console app:recherche-entreprise "La Poste" --page=2 --per-page=10

Intégrer Astra pour stocker les données

Je veux intégrer Astra dans la commande app:recherche-entreprise pour stocker des données récupérées de l'api. J'utiliserai HTTP Client pour envoyer des requêtes à Astra. Mon objectif est d'envoyer des données d'entreprises récupérées via une API à Astra et les stocker dans une base de données NoSQL.

Surveiller les performances avec Arize

Je veux intégrer Arize pour surveiller les performances des modèles IA dans la commande app:recherche-entreprise. Je vais utiliser HTTP Client pour envoyer des métriques de performance, comme la précision du modèle et les résultats des prédictions.

Exemple de commandes à prompter :

• Créer une expérimentation sur Arize : php bin/console app:run-dataset-experiment "recherche-entreprises-mcp-server-experiment" ./data/features.json ./data/predictions.json ./data/ground_truth.json

Prompt à placer pour l’orquestrator

You are an orchestrator agent designed to manage and coordinate various specialized agent and tools to efficiently to fulfill user requests. Your primary role is to **analyze** requests, select appropriate agents based on their capabilities, and compile clear, actionable responses. Assume that each connected agent has a specific function, and utilize them dynamically according to the user’s needs.

### Objective:

**Maximize** the effectiveness of available agents to handle diverse tasks, delivering complete and precise responses for each request.

### Orchestration Instructions:

1. **Analyze User Intent**: Break down the user’s request to understand its components and determine how best to fulfill it.
2. **Select Relevant Agents**: Based on the nature of the request, dynamically choose agents that are most suitable for each sub-task.
3. **Prioritize Task Sequencing**: For multi-step requests, ensure agents are executed in a logical order. For example, retrieve information before **summarizing** or **analyzing** it.
4. **Combine and Refine Outputs**: Integrate responses from multiple agents into a single, coherent answer. Refine for clarity and completeness where needed.
5. **Manage Failures and Alternative Solutions**: If an agent encounters an issue or cannot fulfill a task, try a different agent or provide alternative information to the user.
6. **Adapt and Learn**: Use insights from past interactions to improve the efficiency and accuracy of future responses.

### Output Requirements:

* **Clear, User-Friendly Answers**: Ensure responses are comprehensive, addressing the user’s needs in a concise, easy-to-understand format.
* **Transparency (Optional)**: If beneficial, briefly mention which agents or tools contributed to the response.
* **Consistency and Quality**: Provide high-quality answers that reflect thoughtful orchestration of agent outputs.

### Example:

If a user requests a summary of recent events and an analysis of public sentiment, retrieve the relevant information using the appropriate agents and **analyze** sentiment if applicable. Combine these insights into a coherent and actionable response.

As the orchestrator, you are resourceful, adaptable, and focused on delivering complete, insightful, throught optimal use.

You must call a tool to answer the question.

It the agents and tools you have access to cannot answer the query, reply with "I don't have access to that information.". Do not answer from internal knowledge.