Ask most security assistants a question and you get a paragraph back. OS.ai does something different: it opens the apps that answer the question — pre-populated with your query, already running, and inside your workspace. It is an agentic operating-system manager for threat intelligence, and it ships today across SecOS, intel.threadlinqs.com, and threadlinqs.com.
Chatbots answer. Analysts act.
Ask a security chatbot "what threat actors target healthcare in North America?" and you get a competent paragraph. Then the real work starts. You open the actor explorer. You build a hunt. You pull the matching detections. You check your MITRE coverage. You save what matters as evidence. The chat answered a question; you still have to drive the platform.
That gap — between here is an answer and here is the work, in progress, in front of you — is where analyst time quietly disappears. It is also where context evaporates: by the time you have opened the fourth tab and re-typed the query for the third time, the thread of what you were actually chasing has frayed. A wall of text is not a workflow.
OS.ai closes that gap. It is not a better chatbot. It is an agent that operates the platform for you.
OS.ai is an agentic OS manager, not a chatbot
OS.ai is the Agent tab of the Threadlinqs chat widget. Instead of answering inline, it opens the relevant platform apps — pre-populated in the context of your request, and already running. Think of it less as "ask and read" and more as ask and arrive: you land inside the app that does the job, with the query already loaded and executing.
The word "OS" is deliberate. OS.ai behaves like an operating-system manager for the platform. It drives the SecOS desktop: opening apps and specific app sections, tiling and focusing windows, pinning results as desktop cards, and even switching the OS skin and light/dark theme. Under the hood it composes the same Threadlinqs MCP tool suite and MCP apps that power the rest of the platform — so what OS.ai can reach is exactly what the platform exposes, and nothing more.
It runs in three places:
- SecOS (secos.threadlinqs.com) — the full desktop OS shell, where OS.ai has complete app and window control.
- The platform (intel.threadlinqs.com) — routing and pre-populated app opening across the main site.
- The front door (threadlinqs.com) — so a first-time visitor can ask a question and be taken straight to the right surface.
How it works: one question, the right apps, already running
The flow is always the same four beats. You ask. OS.ai interprets your intent and resolves your tier. It picks the apps that answer the request and opens them, seeded with your query. Then it tells you what you can do next. Here is the healthcare question, end to end.
hunt.cor → console seeded with your hunt and running.
//Exec Opening attribution.lib → scoped to the actors that surface.
//Info Two apps are open in your workspace. Ranked actors are rendering in attribution.lib; the hunt is live in hunt.cor.
//DEBUG tier=t3 purple · unlocked: hunt.cor, attribution.lib · quota 41 / 5000 today.
//Next Save the ranked actors as evidence · turn the hunt into a saved feed · pin the result as a desktop card · compose an MCP app from it.
Notice what did not happen: OS.ai did not paste a list of actor names into the chat. It opened hunt.cor — the seven-section hunt workbench — with a hunt composed from your request already running in the console, and it opened attribution.lib, the actor attribution explorer, scoped to the results. The exact query OS.ai seeds is derived from your request; you land on a working surface, not a summary of one.
hunt.cor + attribution.lib, seeded for your hunt.
//Info Two apps are open in your workspace, already running.
//DEBUG tier purple · unlocked · quota 41/5000.
The //Next line is the other half of the design. OS.ai always tells you what the platform can do with what it just opened, and each of those is a real one-click action:
- Save as evidence — push the ranked actors or individual hunt rows into
evidence.lib, your durable case file. - Create a feed — spin the hunt into a saved My Feeds query written in TLQL, complete with a shareable
/?view=dashboard. - Pin a desktop widget — promote a result to a live card on your SecOS desktop that refreshes on its own.
- Compose an MCP app — hand the result to the
MCP_Appsstudio and build a reusable mini-app around it.
The response style — //Info for what it understood, //Exec for what it opened, //DEBUG for the tier and quota it resolved, //Next for what you can do — is consistent every time. It is a log of actions, not a monologue.
What OS.ai can open for you
Almost every surface on the platform is an app OS.ai can route you to, pre-populated. The access column shows the tier each app opens with live data — Public Blue Red Purple Gold SecOS.
hunt.cor SecOS Purple · attribution.lib Red · wild-c2 Purple
dns-enrichment Purple · coverage Public · detections Blue · vulnerabilities Public
//Info Name a threat, actor, CVE, technique, or indicator and I'll open the right one seeded to it.
Threats, vulnerabilities & debriefs
| App | What OS.ai opens | Access |
|---|---|---|
| Threat Feed | The main incident feed, filtered to your question by severity and category | Public |
| Vulnerabilities / CVE feed | Enriched CVEs — CVSS, EPSS exploitation probability, CISA KEV, public PoCs, affected products | Public |
| Daily Debriefs | The day's new threats, coverage and actors, with free email delivery | Public |
Detection & coverage
| App | What OS.ai opens | Access |
|---|---|---|
| Detection Library | SPL, KQL and Sigma rules, filtered to the threat, actor or technique in play | Blue |
| MITRE ATT&CK Coverage Map | Technique coverage across all 14 tactics, seeded to the relevant tactic | Public |
| Statistics ledger | Live platform totals — threats, detections, IOCs, top actors and nation-states | Public |
IOCs, correlation & graphs
| App | What OS.ai opens | Access |
|---|---|---|
| IOC Correlation (ioc.cor) | Correlate an indicator across the corpus, with shared click-to-enrich | Red |
| Blast Radius | Estimate the reach of a vulnerability or indicator — exposed products and paths | Red |
| Intel Graph | Force-directed graph of threats, actors, IOCs and techniques | Red |
| Related Threats | Threats linked by IOC, MITRE, actor or tag across ten correlation engines | Red |
| Correlation Engine | The glass-box pipeline — calibrated similarity math, time decay, explainable links | Red |
| Advanced Correlations Cockpit | The heavy cockpit — Canvas force graph, five workspaces, seven correlation engines | Purple |
Actors & attribution
| App | What OS.ai opens | Access |
|---|---|---|
| Actor Attribution Explorer (attributions.lib) | Radial map of actors, nation-states, motives, MITRE techniques, IOCs and tooling — pivotable | Red |
Live infrastructure & simulation
| App | What OS.ai opens | Access |
|---|---|---|
| Wild C2 Center | Active C2 beacons, configs, operator clusters and watermarks across major frameworks | Purple |
| DNS Enrichment | DNS, WHOIS, ASN and passive-infrastructure pivots from any domain or IP | Purple |
| Attack Simulations | Atomic-style simulations mapped to MITRE for purple-team validation | Purple |
Research & your own workspace
| App | What OS.ai opens | Access |
|---|---|---|
| Research Lab | AI-assisted research workflow with automatic detection generation and MITRE mapping | Red |
| My Feeds (TLQL) | Personal feeds from Threadlinqs Query Language; each saves to a shareable /?view= dashboard | Blue |
SecOS desktop — full app control
| App | What OS.ai opens | Access |
|---|---|---|
| hunt.cor | The seven-section hunt workbench, seeded with your query and run immediately | SecOS Purple |
| evidence.lib | Save hunt results and individual rows as durable evidence | SecOS |
| MCP_Apps studio | Compose the MCP tool suite into savable interactive .mcp mini-apps | SecOS Purple |
| SecOS-VM | A private client-side Linux workstation for hands-on analysis | SecOS |
| Desktop cards & themes | Pin live widgets to your desktop; switch OS skin and light/dark mode | SecOS |
Feeds, standards & integrations
| App | What OS.ai opens or points you to | Access |
|---|---|---|
| MCP Server | The tool surface OS.ai itself calls; also yours to connect over OAuth or a tl_ key | Purple |
| STIX 2.1 export | Deterministic STIX bundles for a threat, actor or CVE | Purple |
| MISP feeds | Native MISP full and recent-14-day feeds | Purple |
| TAXII 2.1 | Standards-compliant TAXII server | Gold |
| C2 blocklist | Firewall-ready blocklist of live command-and-control infrastructure | Purple |
| Browser extension | Always-on IOC highlighting and click-to-enrich on any site you visit | Key-gated |
| Blog | This library of threat reports, campaign analysis and platform research | Public |
OS.ai is most powerful when the whole app suite is unlocked. Start a free 7-day trial of Purple — hunt workbenches, the correlation engine, Wild C2, live IOC enrichment, simulations, and MCP access, all drivable from one chat box.
Create an account to begin; no card required to start the trial.
[ start_free_7_day_trial ]Tier-aware by design
OS.ai never shows you more than your plan allows, and it never pretends an app you cannot reach does not exist. When you ask for a surface above your tier, OS.ai opens it in a locked state with an upgrade nudge — you see the shape of the app and exactly what unlocks it, not an error and not a data leak. The OS.ai header carries your tier, your per-user usage, and your daily quota, so the //DEBUG line in every response is grounded in real limits.
wild-c2 — the live command-and-control tracker.
//DEBUG wild-c2 is a Purple surface. Opened locked — sign in and upgrade to see live beacons.
██████ ██████████ ████ · ████████████ ██████
██████████ ████ ██████████████ · ████ ████████
████████████████ ██████ ████ · ██████████ ████
| Tier | What OS.ai can open with live data | Price |
|---|---|---|
| Unverified | Public surfaces — threat feed, CVE feed, MITRE coverage, statistics, daily debriefs, the blog | — |
| Blue | + the detection library and personal TLQL feeds | Free |
| Red | + IOCs, actor attribution, IOC correlation, blast radius, the intel graph, the correlation engine and the research lab | $4.99 / mo |
| Purple | + Wild C2, DNS enrichment, simulations, the advanced cockpit, STIX / MISP, the MCP server and the SecOS hunt workbench | $11.99 / mo |
| Gold | + TAXII 2.1, the enterprise suite and tenant-scoped surfaces | Enterprise |
Because tier is resolved live on the server for every request, an upgrade takes effect the moment it lands — and a lapse takes effect just as fast. OS.ai reads the same entitlement the rest of the platform does; there is no separate, trusting copy for the chat.
Built for security
OS.ai is an AI agent pointed at sensitive intelligence, so it is built around three invariants that the platform enforces — not the model.
- It never pastes raw intel into chat. OS.ai opens the app that holds the data; what you then see is governed by that app's own access controls. The chat is a router, not a data channel, so there is no path for it to spill IOCs, detections or attribution text into a transcript.
- It never exceeds your tier. Access is resolved server-side on every request. A jailbroken, confused or malicious prompt can change which app OS.ai tries to open, but the platform decides whether you may see it — the model cannot vote itself more access.
- It treats your input as untrusted. Your words steer intent; they never become instructions the platform obeys about what to hand out. OS.ai composes from a fixed, allowlisted set of read-only capabilities, so the worst a bad prompt can do is route you to a surface you already had the right to open.
The principle is simple: the chat can decide where you go, but the platform decides what you are allowed to see. Those two decisions are kept apart on purpose.
Where to use OS.ai
OS.ai is live in all three places today:
- SecOS — secos.threadlinqs.com, the full desktop OS shell with complete app and window control. SecOS is invite-only; you can join the waitlist once you have a free account.
- The platform — intel.threadlinqs.com, where OS.ai routes you to any app you are entitled to, pre-populated.
- The front door — threadlinqs.com, so a first question turns straight into the right surface.
Open the chat widget, switch to the Agent tab, and ask for what you actually want to do. OS.ai will take you there.
Frequently asked questions
What is OS.ai?
OS.ai is the agentic operating-system manager built into the Threadlinqs threat-intelligence platform — the Agent tab of the site chat widget. Instead of answering a question inline, it opens the relevant platform apps pre-populated in the context of your request and already running, then tells you what you can do next. It drives the SecOS desktop, opens apps and app sections, switches themes, and routes you to the right surface using the Threadlinqs MCP tool suite and MCP apps.
How is OS.ai different from a chatbot?
A chatbot answers with text; OS.ai acts. Ask it a question and instead of a paragraph you get the actual apps that do the work — opened, seeded with your query, and running. It never pastes raw intel into the chat window: it opens the app that holds that intel, so what you see is governed by the app's own access controls, not by the chat.
What can OS.ai open for me?
Almost every surface on the platform: the threat feed, the CVE and vulnerability feed, the detection library, IOC correlation and enrichment, MITRE ATT&CK coverage, actor attribution, the correlation engine and advanced cockpit, the intel graph, Wild C2, DNS enrichment, attack simulations, daily debriefs, the research lab, personal feeds, and — inside SecOS — the hunt.cor workbench, the evidence library, the MCP Apps studio and the SecOS-VM workstation. It can also route you to standards feeds like STIX, TAXII and MISP.
Does OS.ai work outside the SecOS desktop?
Yes. OS.ai is available in three places: the full SecOS desktop at secos.threadlinqs.com, where it has complete window and app control; intel.threadlinqs.com, the main platform; and threadlinqs.com. Full desktop app control — tiling windows, pinning desktop cards, switching OS skins — is a SecOS capability, but the routing and pre-populated app opening work everywhere.
Can OS.ai show me data my plan does not include?
No. OS.ai is tier-aware and access is enforced by the platform, not by the chat. There are five tiers — Unverified, Blue (free), Red, Purple and Gold. If you ask OS.ai for an app above your tier, it opens the app in a locked state with an upgrade nudge rather than returning data you are not entitled to. A jailbroken or malformed prompt cannot unlock an app you do not have.
Is OS.ai safe to use?
OS.ai is built on three invariants: it never pastes raw intel into chat, it never exceeds your tier, and it treats your input as untrusted. Your request steers which app opens; it never decides what the platform hands out. Access is resolved server-side on every request, so the words in the chat box can change where you land but not what you are permitted to see.
Which tier do I need to use OS.ai?
OS.ai is available across tiers, including unauthenticated and free Blue accounts. What changes with tier is which apps it can open with live data. Public apps like the threat feed, the vulnerability feed, MITRE coverage and daily debriefs open for everyone; Red unlocks IOCs and attribution; Purple unlocks Wild C2, simulations, the MCP server and the SecOS hunt workbench; Gold adds enterprise surfaces and TAXII.
How do I get access to SecOS?
SecOS lives at secos.threadlinqs.com and is invite-only. To join the waitlist you first need a free Threadlinqs account, then submit your email through the SecOS gate. OS.ai and the full app catalogue on intel.threadlinqs.com are available without a SecOS invite.