Workbench tools fall into five categories, each serving a different purpose.
Templates: Pre-built integrations
Templates are single actions provided by Tines that connect to external services. They're ready-to-use integrations that don't require you to build a story first. Some examples:
Search Jira for issues
Send a Slack message
Query a database
Look up user information in Active Directory
Create a ticket in ServiceNow
Templates are useful for quick, one-off actions you perform frequently. Instead of building a story for every simple task, you enable the relevant template and ask Workbench to use it.
Stories: Your custom workflows
Stories are workflows built in Tines Stories, either by you or by team members who've made them available. When you enable a story for Workbench, you can trigger it from a conversation, passing in parameters and receiving results back.
Any workflow you've built can become a tool Workbench can use. If you have a story that enriches IP addresses with threat intelligence, Workbench can run it. If you have a workflow that provisions user access, Workbench can trigger it.
MCP tools: Remote server integrations
MCP tools connect Workbench to remote servers that follow the Model Context Protocol (MCP) standard. Each server exposes its own set of tools that Workbench can call directly in a conversation.
MCP tools are useful when you need to reach services that aren't covered by Tines Stories templates. If a third-party platform exposes an MCP server, you can connect to it and use its tools without building a story or waiting for a template to become available.
Skills
Skills are reusable instructions that tell Workbench how to complete a specific task. You create a skill manually or import a markdown file from the Skills page for a team, then enable it for a preset from the Skills tab in Workbench. Workbench loads the full instructions when they're relevant to what you've asked for.
Skills are useful when your team completes the same task the same way every time. Rather than repeating those steps in every conversation, you package them once and let Workbench pick them up when they apply.
Internal tools: Built-in capabilities
Internal tools are built directly into Workbench. You don't need to configure credentials or create stories to use them.
Think: A planning scratchpad where Workbench reasons through complex requests before executing them. It helps Workbench handle multi-step tasks more effectively.
Code analyst: Generates and runs Python code, creates visualizations, and performs data analysis. If you need to process data, create charts, or perform calculations, Code analyst handles it.
Formulas docs: Provides current documentation for formulas. When you ask about formula syntax or capabilities, Workbench consults this tool to give you accurate, up-to-date information.
Memory: Stores context about you and your work so Workbench can recall it in future conversations. Workbench builds this automatically as you work. It uses three underlying tools (create memory, search memory, and search past messages) but you don't configure any of these manually.
Web search: Searches the web so Workbench can draw on current information from beyond your tenant. Your tenant owner controls whether web search is available in AI settings, and tenants with egress restrictions configured can't use it.
💡Note
ℹ️Info
In addition to these four tool types, Workbench supports skills. Skills are packages of specialized knowledge and instructions that Workbench loads on demand when a task calls for them. You create and manage skills from the Skills section in your team settings, and you can toggle which skills are available per preset.