[Question] Can quint-code be used by multiple Agents at the same time?

[A4ABATTERY]

i.e.
Agent 1: Works on implementing a frontend code-change
Agent 2: Works on making test cases
Agent 3: Works on implementing backend code-change

[m0n0x41d]

I am uncertain about my understanding of your use case and the challenges you're facing.

Quint Code is designed for utilization by a human operator who is fully attentive to the process.

In version 4, you may employ parallel agents to implement DRR, it must not be a problem

[A4ABATTERY]

For Contex:

• I'm using Google's Antigravity IDE
• Although it's not recommended, I'm trying to set up an automated workflow with agents running in parallel.

I ran into some issues while starting multiple agents:

• I think at some point the states conflicted:
  • i.e. Agent 1 was on Induction
  • i.e. Agent 2 resets to IDLE.
• I also had the MCP server crash with some erroneous errors
  • "Connection closed calling "tools/calls": client is closing: invalid character 'W' looking for beginning of value

- After this error, I needed kill and restart the MCP server.

I think Antigravity has something to do with the issues I'm running into.
I'm going to move to Gemini-CLI or Claude-Code instead.

[A4ABATTERY]

Reading into the changes in v4.1.1 (#22 ); the error above might be resolved.

I'll update and try it out.

[satzynu]

@A4ABATTERY

i was able to stably use with antigravity (consistent tool calling + local workspace directory set automatically for correct .quint ). However not tested for multi agent state change.

mcp command set to --> quint-launcher.sh which calls quint-launcher.py (attached)

quint-launcher.sh
quint-launcher.py

@m0n0x41d -> may be you can use the approach for directory/workspace detection for correct .quint directory in the go binary itself avoiding the need for setting the ENV variable, which allows for global MCP configuration.