Mobile / APK
Map Android and mobile application tasks by artifact, runtime context, and objective.
Describe the task. The router detects the signals, composes the path, and shows exactly why it chose it.
This reference demonstrates the desired interaction. Production logic remains local, deterministic, and backed by tested route definitions.
The core is pre-loaded with curated specialist routes built from security best practices.
Map Android and mobile application tasks by artifact, runtime context, and objective.
Route browser bundles, client-side logic, and obfuscated JavaScript review tasks.
Start with classification before selecting deeper specialist paths.
Organize model-integration review around boundaries, tools, and trust assumptions.
Structure an API-focused review from surface mapping through evidence organization.
The fallback is intentional: weak evidence should not force a specialist route.
No black box.
No invented route.
The engine uses local, inspectable routing rules. If the evidence is weak, it falls back instead of pretending to know.
Each task is evaluated against known dimension patterns using simple regex and heuristics.
Signals contribute evidence. Only when evidence passes a threshold does a route shift from general to specialized.
The route determines the core phases. Matched signals inject specific execution modules into those phases.
The system outputs the workflow. A human (or executing agent) follows it predictably.
The logic for routing is simple, strict, and executes entirely in the browser using pre-compiled TS logic. It requires zero server infrastructure.
INSPECT SOURCE →The routes, signals, and execution modules are defined in JSON, making it trivial to add custom organizational routing rules or replace the dataset completely.
VIEW SCHEMAS →Fork the repository, drop in your own JSON rules, and build your own deterministic execution orchestrator.
VIEW REPOSITORY