Explore seeded Signal Agents before any public publishing exists. Harbor starts the library with standard house models and strategy-focused variants, each defined by universe scope, cadence, and operating posture. No agent carries performance claims until it earns them; every agent must pass the walk-forward gate before capital. Browse here, build in Studio.
Base, Advanced, and Premium stay inside Signal. Quant unlocks the private builder, broader context, and effectively uncapped universe control.
Harbor-built baselines for the library.
The baseline Harbor signal model. Balanced confluence across trend, momentum, volatility, and participation with a broad US equity universe.
Tracks leadership rotation inside semiconductors and AI compute with heavier relative-strength weighting and tighter participation filters.
A tighter basket focused on AI infrastructure names where trend persistence and expansion in participation matter more than broad market breadth.
A steadier strategy surface centered on cybersecurity names, tuned for persistent relative strength and cleaner post-breakout follow-through.
Focused strategy personalities for users comparing different signal styles.
Ranks candidates by persistent cross-sectional strength, then favors names gaining leadership without already exhausting participation.
Looks for clean breakouts with confirmation across participation, volatility compression, and intermediate trend alignment.
A more selective pullback model for names resetting into support after prior leadership, with penalties for broken structure.
Focused on post-earnings drift setups where surprise, volume expansion, and follow-through create repeatable short-cycle momentum.
Use Signal Studio for prompts, workflow editing, spec review, backtests, and private versions. This library should stay focused on browsing.