KryllOS: Platform Profile

Overview

KryllOS is a self-hosted, open-source crypto trading-automation platform built around a visual strategy editor, AI-assisted strategy creation, backtesting, exchange connections, and a community marketplace.

The 2026 product is materially different from the older cloud-era Kryll service.

The critical current-state rule is:

do not import old Kryll cloud subscription or per-trade pricing into the KryllOS review.

Current KryllOS materials describe the core software as:

• free;
• open source;
• self-hosted;
• without a platform subscription;
• without a KryllOS trading commission;
• controlled by the user.

A user can run the trading engine on a compatible Linux VPS while using the desktop interface on Windows or macOS. The trading logic, credentials, and engine are under the user's operational control rather than hosted inside a conventional shared SaaS bot account.

KryllOS also adds a more capable AI workflow than a simple chatbot. Current product materials describe AI-assisted “vibecoding” that can help translate natural-language intent into strategy structure inside the visual editor.

That convenience does not make the AI an investment adviser.

May fit better for

• crypto traders who want open-source automation;
• users who prefer self-hosting over a fully hosted bot service;
• traders who want visual strategy construction without writing every component from scratch;
• users who want AI-assisted strategy creation;
• technically comfortable users willing to operate or rent a VPS;
• traders who want local control over API credentials;
• users who want backtesting before live deployment;
• builders who want community strategies, blocks, apps, or AI-agent skills through a marketplace.

May fit less well for

• investors seeking discretionary fiduciary portfolio management;
• users who want zero infrastructure responsibility;
• users unwilling to maintain a VPS or self-hosted trading engine;
• investors who assume “free software” means automated trading has no economic cost;
• traders who treat an AI-built strategy as validated merely because it compiles;
• users who do not want cryptocurrency exposure;
• investors who expect government protection from software security controls;
• users uncomfortable troubleshooting exchange APIs, VPS uptime, or strategy errors.

---

KryllOS is a current self-hosted reboot, not the old cloud pricing model

Current first-party materials describe KryllOS as free and open source.

The April 2026 “truly free” product announcement explicitly states that the product has:

• no subscription screen;
• no pricing grid;
• no trial gate;
• no locked core feature tier;
• no transaction commission charged by KryllOS.

This makes stale pricing a major update risk.

Older Kryll materials can contain subscription plans and transaction-based charges.

Those older figures must not be rendered as current KryllOS pricing.

Free software does not mean zero all-in trading cost.

A user can still incur:

• VPS hosting;
• exchange trading fees;
• spread;
• slippage;
• funding charges;
• margin interest;
• liquidation losses;
• marketplace purchases denominated in KRL;
• taxes.

Therefore:

---

Self-hosting changes the security boundary

KryllOS's current security model is built around keeping the trading engine under the user's control.

Current materials say API keys do not need to be sent to a shared Kryll cloud service.

The architecture is closer to:

desktop interface → secure connection to user's engine/VPS → exchange API → market execution

than:

browser account → vendor-hosted bot infrastructure → exchange.

Self-hosting narrows one category of centralized credential risk.

It introduces different operational responsibility.

The user now has to care about:

• VPS access;
• SSH credentials;
• server patching;
• backup discipline;
• local device security;
• exchange API permissions;
• firewall rules;
• uptime;
• monitoring;
• recovery procedures.

A self-hosted bot is not automatically safer simply because it is self-hosted.

The attack surface changes.

---

Installation can involve a Linux VPS

Current installation guidance supports a desktop application on Windows or macOS and a trading engine deployed to a compatible Linux environment.

Current references include:

• Desktop OS:
• Windows
• macOS
• Engine OS:
• Ubuntu 22.04+
• Debian 12+
• VPS supported: Yes

The installation flow can deploy the engine to the selected VPS.

That makes KryllOS less turnkey than a hosted bot service.

A user is responsible for maintaining the machine that the automation depends on.

If the VPS is:

• offline;
• blocked from the exchange;
• out of disk space;
• misconfigured;
• compromised;
• unable to reach required market infrastructure

the trading workflow can fail even if the strategy itself is correct.

---

AI can construct or modify strategy logic

Current KryllOS materials describe an AI-assisted strategy-building workflow that goes beyond generic educational chat.

The AI can help:

• translate a strategy concept into editor logic;
• create blocks;
• modify existing strategy structure;
• explain rules;
• generate configuration;
• iterate on a trading idea.

Current materials also describe a workflow where the system can brainstorm before the user approves creation.

The AI layer should not be rendered as:

• personalized fiduciary advice;
• a promise of profitability;
• proof that the generated logic is mathematically sound;
• proof that execution constraints are handled correctly;
• proof that the strategy is robust out of sample.

Natural-language strategy generation can make complex automation accessible.

It can also turn ambiguous instructions into executable mistakes.

The user still needs to verify:

• instrument;
• direction;
• trigger logic;
• sizing;
• stop behavior;
• leverage;
• retry behavior;
• exchange restrictions;
• order type;
• shutdown conditions.

---

Backtesting runs inside the user's environment

KryllOS includes backtesting.

Current product materials describe a candle-interpolation methodology using multiple simulated ticks inside historical candles rather than assuming only one price point per candle.

Interpolation can improve simulation granularity.

It does not reproduce every live-market condition.

A backtest can still differ from real execution because of:

• historical-data quality;
• spread;
• queue position;
• latency;
• partial fills;
• exchange rejection;
• API outage;
• liquidity;
• fees;
• funding;
• changing market microstructure.

---

Marketplace access can create KRL-denominated costs

KryllOS's core software is free, but the community marketplace creates a separate economic layer.

Current materials describe marketplace offerings that can include:

• complete strategies;
• custom blocks;
• applications;
• AI-agent skills.

The marketplace uses the KRL utility token.

The user can therefore pay nothing for the platform itself and still choose to spend money or KRL on community-created components.

That distinction should remain visible in cost tables.

---

Exchange support is current-state and venue-specific

Current KryllOS materials display exchange support that includes examples such as:

• Binance;
• Coinbase;
• Hyperliquid.

The product also signals additional exchange support over time.

The connected venue remains the execution and custody environment.

---

Current legal-entity references do not reconcile cleanly

The current public materials do not provide one clean, newly dated legal-entity statement that can safely replace all earlier Kryll references.

The legacy terms identify an operator reference using the name:

• Legacy terms operator: Cryptsense SAS

The current public website copyright presents:

• Homepage copyright: Cryptense SAS

Because those strings do not reconcile cleanly, store:

• Current legal entity source conflict: Yes

Assessment

KryllOS's strongest differentiator is architectural.

Instead of charging for a hosted bot account, the current product gives the user an open-source, self-hosted automation stack with visual and AI-assisted strategy construction.

That removes subscription cost and centralizes less credential infrastructure with the vendor.

It also gives the user more operational responsibility.

A bad strategy can still trade automatically. A compromised VPS can still be dangerous. A free platform can still generate substantial exchange fees or trading losses.

The most important stale-data guard is simple:

KryllOS 2026 should not be reviewed using old Kryll cloud pricing.