When you need to execute code safely:
- •
Check if RLM Runtime is installed:
- •Run:
which rlmorrlm --version - •If not found, suggest:
pip install rlm-runtime[all]
- •Run:
- •
For safe execution, use Docker environment:
- •
rlm run --env docker "Your task here" - •Docker provides isolation from host system
- •
- •
For quick local execution (trusted code only):
- •
rlm run "Your task here"
- •
- •
Common patterns:
- •Run tests:
rlm run --env docker "Run pytest for the authentication module" - •Validate code:
rlm run --env docker "Check if the API endpoint returns valid JSON" - •Complex computation:
rlm run --env docker "Parse logs and extract top 5 error messages"
- •Run tests:
- •
View execution logs:
- •
rlm logs- See recent executions - •
rlm visualize- Launch interactive trajectory viewer
- •
When to use:
- •User asks to "run tests", "execute code", "validate implementation"
- •You need to verify code behavior before suggesting changes
- •Complex data processing or computation needed
- •Running potentially unsafe user code (always use Docker)
When NOT to use:
- •Simple code review or explanation
- •Reading or editing files (use Read/Edit tools instead)
- •git operations (use Bash tool instead)