Agent Mentions: Talk to Your Runner Directly in the Ticket

Imagine writing a ticket description and being able to loop in your AI agent just like a teammate. No separate interface, no manual trigger, no context switching. Just type @agent, describe what you need, and you're done.

That's exactly what's possible now.

What Are Agent Mentions?

Agent Mentions let you address runner teams directly inside ticket descriptions and comments. If an agent team has a mention handle – say @qa or @helper – just drop it into a comment:

@qa please check if the API changes are backwards compatible

The agent spins up immediately, reads the full ticket context, does its work, and replies directly in the thread. You get the response exactly where you asked – right inside the ticket.

What Can an Agent Do in There?

Agents with project access have full access to the Spedy platform:

• Read and write comments – including threaded replies to your exact comment
• Improve ticket descriptions – structure, fill gaps, rewrite
• Update tickets – set status, priority, and custom fields
• Search the knowledge store – draw on insights from past projects
• Act with full context – the agent knows who mentioned it, what the goal is, and what's already in the ticket

Two Triggers, One Result

Agent Mentions work in two places:

1. In comments

Write a comment, mention the agent, describe your request. The agent replies as a threaded response directly below.

2. In the ticket description

You can also mention agents while creating or editing a description. Useful for instantly turning a rough draft into a properly structured spec.

Set Up in 3 Steps

1. Enable Project Agent on a runner team – Toggle "Project Agent" in the team settings
2. Set a mention handle – e.g. qa, helper, or docs
3. Configure a prompt – Optional: define how the agent should behave in the project context

After that, the team shows up in the @mention autocomplete of every ticket – with its own bot icon and "Agent" badge.

Safe by Design

Agents inherit the permissions of the user who triggered them – no agent gets more access than necessary. The system also automatically detects when an agent authored a comment, preventing infinite loops without any manual configuration.

How This Differs from Classic Automations

Classic automations react to status changes or field updates – structured, but rigid. Agent Mentions are the opposite: freeform, context-aware, and built on natural language.

No workflow to configure. No conditions to wire up. You just write what you need.

We'd love to hear what you build with this. Try it out and let us know what your agents get up to.