Feature Request
Add support for running multiple agents in parallel, where each agent gets its own dedicated iOS Simulator instance managed by RocketSim's CLI.
As agentic development workflows grow, developers increasingly want to fan out work across several agents at once (e.g. best-of-N attempts, parallel feature exploration, regression sweeps). Today, agents can collide on the same Simulator, which leads to flaky automation and unreliable interactions.
Proposed behavior
- The CLI can allocate a unique Simulator per agent (boot/shutdown lifecycle managed automatically).
- Each agent's
rocketsim commands are scoped to its assigned Simulator (so taps, snapshots, recordings, etc. don't interfere with other agents).
- Clean teardown when an agent finishes.
Example usage
rocketsim agent run --parallel 4 --device "iPhone 17 Pro"
Or a session-based model:
rocketsim session start --device "iPhone 17 Pro"
# subsequent commands within this session target the assigned Simulator
rocketsim session end
Why
- Enables true parallel agentic development (e.g. multiple Cursor / Claude Code agents at once)
- Avoids race conditions on a shared Simulator
- Scales naturally with best-of-N patterns and parallel CI runs
Requested by Eslam via X.
Feature Request
Add support for running multiple agents in parallel, where each agent gets its own dedicated iOS Simulator instance managed by RocketSim's CLI.
As agentic development workflows grow, developers increasingly want to fan out work across several agents at once (e.g. best-of-N attempts, parallel feature exploration, regression sweeps). Today, agents can collide on the same Simulator, which leads to flaky automation and unreliable interactions.
Proposed behavior
rocketsimcommands are scoped to its assigned Simulator (so taps, snapshots, recordings, etc. don't interfere with other agents).Example usage
rocketsim agent run --parallel 4 --device "iPhone 17 Pro"Or a session-based model:
Why
Requested by Eslam via X.