Back to rankings

Mistral Vibe: Slash Commands While Agent Runs

Mistral VibeView original changelog

Mistral Vibe now lets developers open the slash-command picker and adjust setting pickers even while the agent or a running command is still executing, removing the need to wait for a turn to finish before issuing the next instruction. The change is part of the v2.24.3 release, which also renamed the CLI process to "Vibe CLI" with a visible PID and switched model displays to human-readable names. Together these changes make long-running agent sessions feel more responsive and interruptible for everyday users.

Key Takeaways

  • Slash commands now work mid-turn, letting users open the command or settings picker while the agent is still actively running instead of waiting for it to finish.
  • The change directly targets long-running agent sessions, which have become more common since Mistral Vibe introduced subagents and skills in its 2.0 release.
  • Mistral Vibe's CLI process is now explicitly labeled "Vibe CLI" with its PID shown in the status bar, easing session identification for scripting and multi-session workflows.
  • Model pickers now show human-readable display names rather than raw config aliases, reducing confusion when choosing or switching models.
  • Connector bootstrap failures are now surfaced to the user instead of silently appearing as "no connectors," making broken MCP setups easier to diagnose.
  • The release is a maintenance-forward point update (v2.24.3) following the more feature-heavy v2.24.0 and v2.24.2 releases, focused on session responsiveness rather than new capabilities.

Slash Commands No Longer Wait for the Agent

Mistral Vibe shipped version 2.24.3 on August 20, 2026, and the headline change is a workflow fix that long-time users of the CLI will feel immediately: the slash-command picker and setting pickers can now be opened and used while the agent or a command is still running. Previously, invoking /model, /config, or any skill-provided slash command required waiting for the current turn to finish — an interruption-unfriendly design for a tool built around long, multi-step agent runs.

With this change, developers can queue up a setting change, switch models, or trigger a skill-based command mid-turn without breaking their flow or forcing the agent to stop and restart. This matters most in sessions where an agent is working through a long task — refactoring a large codebase, running a multi-step build, or chaining subagent calls — where previously a user who wanted to tweak configuration had to either interrupt the agent entirely or wait it out.

Why It Matters

Mistral Vibe has increasingly leaned into longer-running, more autonomous agent sessions since the 2.0 release introduced subagents and skills. As sessions get longer, the cost of not being able to adjust settings mid-run grows. This release closes that gap, making the CLI feel more like a live, steerable environment rather than a strict turn-based prompt loop.

Supporting Changes

Alongside the headline feature, Mistral Vibe made two smaller adjustments in this release. The CLI process itself is now named "Vibe CLI" and the bottom status bar displays its process ID (PID), making it easier to identify and manage running Vibe sessions from outside the tool — useful in scripting or multi-session setups. Separately, the /model picker and related displays now show a model's human-readable display name instead of its raw configuration alias, reducing confusion when multiple aliases point to similar underlying models.

Bug Fixes

The release also fixed one notable issue: failures during connector bootstrap are now surfaced directly to the user instead of silently reporting "no connectors," making it easier to diagnose broken MCP or connector setups.