Skip to content

Modulatio

MODULATIO

SYSTEM READY.

Modulate across models, agents, and registers.

A multi-model agent framework with skills-first orchestration, goal-state compression for long-horizon work, and QC that authors fixes instead of abandoning rejected tasks.

WHAT IS MODULATIO

ῥεῖ πάντα — everything flows.

Modulatio is a framework for building multi-agent systems where the hard part — choosing which model does what, when to compress the working context, how to recover from a failed step — happens structurally instead of by prompt-engineering accident.

It externalizes the deliberation-and-acting dialectic across three role tiers: a Leader that reasons, Producers that collaborate-and-act with skill-routed work, and Quality Control that verifies by execution and authors fixes when producers can’t converge.

HOW MODULATIO HELPS YOU

Finish sooner. Independent work runs in parallel. Agents take on tasks asynchronously, so a multi-step project moves at the pace of its longest thread — not the sum of every step run end to end.

Spend far less. Cheap models do the heavy lifting; a stronger model’s judgment catches the errors and sharpens the result, only where it earns its cost. You pay like a budget model and ship like a frontier one.

Think wider. Compose models from different providers and training lineages, and the team inherits their differences — where one model is blind, another sees. Variety of approach becomes a quality edge instead of a single model’s blind spots, repeated.

Run anywhere. No preference for local or cloud — lean on local models to cut cost further, mix in cloud where it pays, and keep working when the connection drops. The factory doesn’t stop at the edge of the network.

One team, one sandbox. Modulatio is two things at once: a coding harness for hands-on work, and a long-horizon, self-looping digital factory staffed by a cast of models you design. Not many copies of one model, not sandboxed agents passing notes over a wall — your models, harnessed into a single cooperative team inside one sandbox.

WHO IT’S FOR

Researchers

Long-horizon work where context-window economics matter. Goal-state compression keeps the system on-task past the point where naive history would saturate the window.

Builders

Production agent systems where rejected tasks need to be recovered, not left dead. QC-as-fixer turns the verifier into a producer-of-fixes.

Pragmatists

Model-agnostic by design. Tested across sparse-MoE, dense, and reasoning models with explicit notes on which slots favor which architectures.

INSTALL

$ pip install modulatio$ modulatio setup$ 

Requires Python 3.12+. See the install guide for the full walkthrough.