Assistant's Tools Overview

Assistant's tools are powerful enhancements that bring completely new capabilities to AI/Run CodeMie. This page provides a comprehensive overview of all available tools.

---

Available Tools

Tool	Description
Keycloak	Access management and authentication tool that allows users to leverage AI/Run CodeMie for project access control
SonarQube	Code quality analysis and static code review tool for comprehensive codebase analysis
SQL	Database operations tool for querying and managing SQL databases
Telegram	Send notifications to Telegram channels and bots
Azure	Microsoft Azure cloud platform integration for cloud resource management
GCP	Google Cloud Platform integration for GCP resource operations
AWS	Amazon Web Services integration for AWS cloud operations
Kubernetes	Kubernetes cluster management tool for container orchestration
Webhook	HTTP webhook integrations for event-driven automatization
Elasticsearch	Search and data analysis platform for analytical queries and data access
Codebase Tools	Comprehensive code analysis and management with repo browsing, code search, and file operations
Email Sender	Send email notifications from assistants
OpenAPI	Integrate with any third-party tools supporting the OpenAPI standard
Quality Assurance	Test management with Zephyr Scale (formerly TM4J) and Zephyr Squad for Jira environments
ServiceNow	ITSM operations including incident management, change requests, and CMDB operations
Scheduler	Task scheduling and automation (Admin role only)
Plugin	Custom plugin integrations for extending assistant capabilities (e.g., file system)
Git AzureDevOps	Azure DevOps integration for Wiki, Work Items, and Test Plans management
Adding MCP Server	Add and configure Model Context Protocol (MCP) servers
Using MCP Tools	Use MCP tools in assistants for extended capabilities

---

Smart Tool Selection

Smart Tool Selection is an intelligent feature that automatically selects relevant tools for your assistant based on semantic similarity, configured credentials, and permissions.

How to Enable

1. Navigate to your assistant's Create or Edit page
2. Locate the Smart Tool Selection toggle
3. Enable or disable based on your needs

Behavior

When No Tools Are Pre-selected

The system performs semantic analysis across all registered tools and automatically selects based on:

• Semantic relevance to your assistant's purpose and conversations
• Available credentials and proper authentication
• User permissions and access rights

When Tools Are Already Selected

• Smart selection is limited to your pre-selected tools only
• Prevents context overload by maintaining a focused tool set
• Provides intelligent selection within your curated tool collection

Important Considerations

Workflow Operations

• Workflows always use the legacy/traditional approach
• The Smart Tool Selection toggle does not affect workflow behavior
• This ensures workflow consistency and predictable execution

Context Management

Smart selection prevents LLM context overload and maintains optimal performance by limiting tool scope.

When to Use Smart Tool Selection

Enable Smart Tool Selection when:

• You want automatic, intelligent tool discovery
• Your assistant needs dynamic tool access based on conversation context
• You prefer minimal manual tool configuration

Keep it Disabled when:

• You need precise control over tool availability
• You're working with sensitive tools requiring explicit selection
• You prefer the traditional manual approach

---

Tool Integration Requirements

The table below shows which tools require integration setup and the integration name to use:

Tool Name	Integration Required	Integration Name	Requires Project Admin
Project Management => Jira	✅ Yes	Jira	❌ No
Git	✅ Yes	Git	❌ No
VCS => GitHub / GitLab / AzureDevOps	✅ Yes	Git	❌ No
Project Management => Confluence	✅ Yes	Confluence	❌ No
Cloud => Kubernetes	✅ Yes	Kubernetes	❌ No
Cloud => AWS	✅ Yes	AWS	❌ No
Cloud => GCP	✅ Yes	GCP	❌ No
Cloud => Azure	✅ Yes	Azure	❌ No
Access Management => Keycloak	✅ Yes	Keycloak	❌ No
Cloud => AzureDevOps	✅ Yes	AzureDevOps	❌ No
Azure DevOps Wiki	✅ Yes	AzureDevOps	❌ No
Azure DevOps Work Item	✅ Yes	AzureDevOps	❌ No
Azure DevOps Test Plan	✅ Yes	AzureDevOps	❌ No
Data Management => Elasticsearch	✅ Yes	Elastic	❌ No
OpenAPI	✅ Yes	OpenAPI	❌ No
Plugin	✅ Yes	Plugin	❌ No
FileSystem => Code Interpreter	✅ Yes	FileSystem	❌ No
FileSystem => Generate Image	❌ No	---	❌ No
Notification => Email	✅ Yes	Email	❌ No
Codebase Tools => Sonar	✅ Yes	Sonar	❌ No
Codebase Tools => Get Repo Tree (Experimental)	❌ No	--- *	❌ No
Codebase Tools => Search Code (Experimental)	❌ No	--- *	❌ No
Codebase Tools => Read Files Content	❌ No	--- *	❌ No
Codebase Tools => Read Files Content with Summary	❌ No	--- *	❌ No
Data Management => SQL	✅ Yes	SQL	❌ No
Notification => Telegram	✅ Yes	Telegram	❌ No
Webhook	✅ Yes	Webhook	✅ Yes
Quality Assurance => ZephyrCloud	✅ Yes	ZephyrCloud	❌ No
Quality Assurance => ZephyrSquad	✅ Yes	ZephyrSquad	❌ No
Research => Google Search	❌ No	---	❌ No
Research => Google Places	❌ No	---	❌ No
Research => Wikipedia	❌ No	---	❌ No
Research => Tavily Search	❌ No	---	❌ No
Research => Web Scraper	❌ No	---	❌ No
Scheduler	✅ Yes	Scheduler	✅ Yes

* For these Codebase tools, Git data source should be added as additional context.

Git vs VCS

By choosing Git tool only: The operations listed above are available. Git data source must be added as additional context.

By choosing VCS tool: All operations are available without needing additional Git data source context.

---

MCP (Model Context Protocol)

AI/Run CodeMie supports MCP servers for extending assistant capabilities.

Topic	Description
Adding an MCP Server	Learn how to add and configure MCP servers
Using MCP Tools in Assistants	Learn how to use MCP tools within your assistants

---

Need Help?

For detailed instructions on configuring specific tools, click on the tool name links in the tables above to view the dedicated guide for each tool.