Mastra Studio adds Thread View to read agent conversations next to traces
Mastra has added Thread View to Mastra Studio, letting developers read an agent's conversation next to the trace that produced it. According to Mastra, users can open an agent trace, see its messages beside the steps behind them, and then open the full thread to follow the whole conversation in order.
- Mastra Studio now shows agent messages beside the trace steps that produced them.
- Users can open the full thread to see the whole conversation in order.
- Agents already using memory and MastraStorageExporter need no additional setup.
- The feature requires [email protected] or later and appears only for traces that belong to a thread.
- Hovering over a message and selecting Highlight spans fades the rest of the span tree.
Starting from the conversation, not the tree
Agent activity is usually traced as a tree of nested steps, model calls and timestamps. Mastra says that although a tree view is precise, it is difficult to interpret, because you have to reconstruct the conversation to understand what happened.
Thread View reverses that approach. According to Mastra, you start from the messages, and when something looks wrong, you dig into the steps behind it. The company argues that this makes traces easier to read, follow and navigate.
The problem grows with session length. Mastra notes that a single session can span dozens of messages, and the longer an agent runs, the harder it is to find the right step. When debugging, developers want to follow what the user asked, what the agent decided and which tools it called. Thread View is meant to let them inspect a trace in the context of the whole conversation.
Mastra also points to a second audience: subject-matter experts who need to review an agent's behavior but did not build the agent.
How to set it up
Teams whose agents already use memory and persist traces with MastraStorageExporter need no additional setup, according to Mastra. Thread View works with their existing stored traces.
For new projects, Mastra lists an install command for the mastra@latest, @mastra/core, @mastra/observability, @mastra/memory and @mastra/libsql packages. The feature requires [email protected] or later, added in PR #25812. Thread View appears only for agent traces that belong to a thread.
The example in the announcement does three things:
- Defines an agent with memory enabled.
- Configures a Mastra instance with LibSQL storage and an Observability setup that uses MastraStorageExporter.
- Calls the agent with both a thread and a resource identifier.
Each call becomes one turn in the thread and one trace in Studio.
Using Thread View in Studio
After setup, developers run mastra dev, open Traces in Studio and select an agent run. The conversation for that turn appears beside the span tree.
Hovering over any message and selecting Highlight spans shows the steps behind that message, with the rest of the tree faded. Selecting Open full thread displays the whole conversation.
Mastra points readers to its documentation for full configuration options, including pages on Studio observability, tracing and memory.
What it means for teams building agents
For developers already storing traces with memory enabled, the change requires no new configuration beyond a compatible Mastra version. Debugging can begin from what the user and agent actually said rather than from a nested span tree. Because the feature is tied to threads, agents called without a thread will not show it.
The focus on reviewers who did not build the agent suggests the feature is also aimed at wider team workflows, where people checking agent behavior may be less familiar with the tracing internals.
