Runtime and Instrumentation Engine
Turning user code into safe, deterministic trace events.
“A traditional debugger observes a program. AlgoLens transforms execution so each step can be understood, explained, and replayed.”
A safe bridge from languages to universal events
The runtime layer answers the hardest practical question: how does raw user code become a Universal Trace? Pure runtime observation is reliable but language-dependent and semantically limited. A custom interpreter provides maximum control but is slow and expensive to maintain. AlgoLens chooses a hybrid strategy: parse the source, instrument the AST, run the transformed program in a sandbox, generate events, and feed them into the trace engine.
AST-level instrumentation lets AlgoLens work with functions, calls, conditions, loops, expressions, returns, and mutations as concepts instead of brittle source text. The platform can inject hooks before and after statements, loops, functions, branches, exceptions, and execution boundaries. Plugins such as memory profilers and complexity analyzers can subscribe to these hooks without knowing each other.
Because executing user code is dangerous, sandboxing is non-negotiable. User programs must have no filesystem, socket, process, environment, disk, secret, or API-key access by default. Every execution should enforce maximum time, memory, and step limits. Even exceptions and limits become trace events, transforming failure into a pedagogical object rather than an opaque crash.
Source Code ↓ Parser ↓ AST ↓ Instrumentation ↓ Sandboxed Runtime ↓ Event Generation ↓ Universal Trace Engine