Browser-native processor learningWrite it. Run it.
Write it. Run it.
See the machine think.
Ark brings source, machine state, guided practice, and a verification-first AI into one focused workspace—from a first loop to a custom architecture.
- No install
- Guest simulator
- Real execution
sum-loop.asmArk MIPSmain:
li $t0, 1
li $t1, 10
move $t2, $zero
loop:
add $t2, $t2, $t0
addi $t0, $t0, 1
ble $t0, $t1, loopFour learning paths
The right machine for every layer of understanding.
From source to understanding
A complete feedback loop, not another code editor.
Every surface answers the same question: what did this instruction do to the machine?
- 01
Write
Use processor-aware source, examples, and instruction references.
- 02
Execute
Animate control flow or run directly to the result.
- 03
Inspect
Watch registers, memory, output, and display state change.
- 04
Practice
Submit the editor’s real code to architecture-specific tests.
- 05
Ask
Let Ark AI explain or verify a repair in a bounded sandbox.
Built for learning
Powerful enough to explore. Clear enough to teach.
Architecture-aware
The editor, debugger, challenges, and AI share one explicit processor contract.
State in context
Source lines stay connected to registers, machine words, memory segments, and output.
Focused challenges
A dedicated workspace, progressive hints, and real state assertions turn examples into deliberate practice.
Verification-first AI
Code-changing answers execute in the selected processor sandbox before they reach the editor.
Browser-native
No toolchain setup is required for the simulator; students can begin as guests.
A visible trust boundary
AI access is gated, credentials stay with Firebase, and model code never receives them.