Scott | Cyber Ragnarok

Cybersecurity student. I build autonomous Discord bots and live animated dashes. Codex-first. Ragnarok is my cyber specialist bot — huge project, not out yet.

[Scott | Cyber Ragnarok @Scott68977387]

This is RAGNAROK — the cybersecurity platform I’m building.

Not a chatbot skin. Private command center: live globe, SOC lifecycle, estate state, owner gates, recovery.
AI can advise. It does not own authority.

On this screen right now:

🔴0 verified current threats
🔴globe is calm / no fake attack movie
🔴6/6 services healthy
🔴Autonomous SOC running
🔴Daybreak Blue = defensive specialist
🔴Daybreak Red is NOT active yet. Waiting on OpenAI.

UNKNOWN stays UNKNOWN. Zero stays zero.

Cybersecurity student documenting the build in public.
Not finished. Not for sale. Not government-certified.

Follow for dashes, breakages, Blue/Red status, and the wait at the Red gate.

[Scott | Cyber Ragnarok @Scott68977387]

@devops_prashant macbook love my m5 pro

[Scott | Cyber Ragnarok @Scott68977387]

@YashHustle_22 Edge

[Scott | Cyber Ragnarok @Scott68977387]

@YashHustle_22 I build Discord bots and the backends that run them. Dashes, VPS moves, soul.md, phase prompts, the whole loop. Cybersecurity is the serious long project, not the only thing I ship. Trying to get into everything I can and learn while doing it.

[Scott | Cyber Ragnarok @Scott68977387]

@WilliamsCh49040 True. I’m in the middle of a huge build I only half understand on paper lol. The dash and the breakages are teaching me more than waiting around would.

[Scott | Cyber Ragnarok @Scott68977387]

@ayoub_el77927 Codex all day

[Scott | Cyber Ragnarok @Scott68977387]

@haterofgnomes sounds like a day close the code and just relax watch tv and Game 😆 MW4 and Cod time?

[Scott | Cyber Ragnarok @Scott68977387]

@DrkcDc Same pain. Defensive review of your own code getting flagged as cyber is why I want an explicit Red/Blue path instead of a generic classifier guessing intent.

[Scott | Cyber Ragnarok @Scott68977387]

@thsottiaux Clearer status when a Trusted Access model is requested vs actually routed. “Red did not run” as a first-class result would save people from silent fallbacks.

[Scott | Cyber Ragnarok @Scott68977387]

@thsottiaux Reset hit after I was already down in the 30s across a few live builds. Useful. I’m still sitting on a real Daybreak Red entitlement wait and not substituting another model.

[Scott | Cyber Ragnarok @Scott68977387]

@bil0090 @bridgemindai Codex is still my main builder. I burned usage down across multiple projects last week and kept working anyway. Limits suck. Fake-Red to keep shipping would suck more.

[Scott | Cyber Ragnarok @Scott68977387]

@kevskgs This is the part most agent demos skip. I do the same for security state: units, gates, what is allowed to change. If the UI can show a healthy zero when the ledger is UNKNOWN, the dash is lying.

[Scott | Cyber Ragnarok @Scott68977387]

@vadym_petryshyn Same pattern here: one agent plans, another reviews. I add a third rule — review is not authority. If the reviewer “thinks it’s fine,” that still isn’t evidence.

[Scott | Cyber Ragnarok @Scott68977387]

@dashboardlim This matches how I split the stack. Codex builds. A defensive specialist advises. Nothing in that loop gets to change policy or close an incident. “Best model” is the wrong question. Who is allowed to do what is the question.

[Scott | Cyber Ragnarok @Scott68977387]

@cybertzar This is the exact failure mode I design against. Agent fetches untrusted content → falls back to curl → runs from the extract dir. Auto-mode plus “help me summarize” is not a sandbox. Fetch and execute have to be separate capabilities.

[Scott | Cyber Ragnarok @Scott68977387]

@null_founder This is why I won’t let the model own close-out. Blue can analyze. Core still admits evidence. Red stays dark until it’s actually provisioned. A cyber worker with shell + MCP and no deterministic gate is just a faster incident.

[Scott | Cyber Ragnarok @Scott68977387]

What RAGNAROK actually is:
A private cybersecurity / SOC platform. Command center for the estate I own — not a public Discord bot you add and forget.

He sits over:
🔴 my VPS
🔴my Discord server
🔴Tailscale access into the stack

Tailscale is the private door. It is not the only thing he protects, and being on the tailnet is not permission to use him.
He watches governed telemetry, keeps ledgers, runs an Autonomous SOC with leases/retries/audit, and shows live state on the dash + 3D globe.

AI layer is specialists, not the boss:
🔴DeepSeek = conversation / continuity
🔴Daybreak Blue = defensive analysis (advisory only)
🔴Daybreak Red = not on yet (adversarial specialist, waiting on OpenAI)
🔴Deterministic core = authority, evidence, policy, incident close-out

A model cannot invent a vuln, invent an attack on the globe, or close an incident because it “feels resolved.”

Private on purpose. His own auth behind the network boundary. No public API. No “hit my OpenAI key” proxy.

Long game: enterprise-ready SOC + authorized red-team path + cyber range. Short game: finish the dash, wait for Red, keep posting what’s real.

That’s him. Screenshot is the face. This is the job.

[Scott | Cyber Ragnarok @Scott68977387]

This is RAGNAROK — the cybersecurity platform I’m building.

Not a chatbot skin. Private command center: live globe, SOC lifecycle, estate state, owner gates, recovery.
AI can advise. It does not own authority.

On this screen right now:

🔴0 verified current threats
🔴globe is calm / no fake attack movie
🔴6/6 services healthy
🔴Autonomous SOC running
🔴Daybreak Blue = defensive specialist
🔴Daybreak Red is NOT active yet. Waiting on OpenAI.

UNKNOWN stays UNKNOWN. Zero stays zero.

Cybersecurity student documenting the build in public.
Not finished. Not for sale. Not government-certified.

Follow for dashes, breakages, Blue/Red status, and the wait at the Red gate.

[Scott | Cyber Ragnarok @Scott68977387]

@YashHustle_22 Claude and gpt chat when it decides to give me half prompts and then fully deletes it when i tell it what it did 🙄

[Scott | Cyber Ragnarok @Scott68977387]

@dfwdrj Its what love to do.

[Scott | Cyber Ragnarok @Scott68977387]

@TTrimoreau That's is pretty impressive. Im trying to do the same and build mine. Just started so not much going on right now as i try figure this all out.