Security systems, creative machines, and useful trouble.

Cybersecurity & AI

Cybersecurity

Security must be built into the system, not sprayed on top after the demo works. No fake production claims. No demo defaults pretending to be customer-ready. No dumping risk onto users because the builder got bored before the hardening work.

The useful version of security is practical: identity, logging, least privilege, detection, response, secure defaults, supportability, threat modeling, remediation, and documentation honest enough that a customer can trust it.

Cybersecurity & AI

AI Security

I build and think about AI systems, prompt injection, model behavior, AI tool intake, enterprise monitoring, SecOps search, classifiers, governance, and safe adoption in large organizations.

AI should be a tool, not a boss, judge, cop, priest, replacement conscience, or hidden policy engine. If a human life, job, freedom, reputation, safety, or future is affected, humans remain responsible.

Security

Secure Architecture

Security work starts with the shape of the system: identity, trust boundaries, data paths, logging, failure modes, and the parts attackers will touch first. A useful architecture makes those things explicit before production pressure turns them into folklore.

• Map trust boundaries before adding new privileges.
• Make logging and evidence part of the launch path.
• Prefer secure defaults that operators can understand and support.

AI Systems

AI Governance

Enterprise AI needs policy, monitoring, intake, review, and accountability in one place. The goal is not to block useful work. The goal is to keep hidden AI behavior from turning into hidden policy.

The durable pattern is a control plane: approved tools, model-use visibility, risk tiers, logging, review paths, and human accountability where consequences matter.

AI Systems

Prompt-Injection Defense

Prompt injection is a systems problem. The model sees text, tools, memory, and user intent inside one context window, so the application has to decide which instructions deserve authority.

• Separate trusted instructions from retrieved or user-supplied text.
• Gate tools with policy outside the model.
• Treat exfiltration attempts as workflow failures, not prompt failures.

AI explainer

How AI Works

Modern AI systems are pattern engines trained on enormous amounts of examples. A language model does not “know” things the way a person knows them. It predicts useful next tokens based on patterns it learned during training, then follows the instructions and context it is given at runtime.

The practical version: a user asks a question, the system builds a prompt, the model generates a response, and surrounding controls decide what tools it can use, what data it can see, what policies it must obey, and what should be blocked or logged.

The parts that matter

• Model: the trained engine that generates, classifies, summarizes, reasons, or transforms content.
• Prompt: the instructions, context, and user request passed into the model.
• Context window: the amount of text or data the model can consider at one time.
• System instructions: higher-priority rules that shape behavior.
• Tools: external capabilities like search, code execution, databases, calendars, tickets, or APIs.
• Guardrails: policy, filtering, validation, logging, and human review around the model.

The blunt truth: AI quality is not just “which model did you pick?” It is the whole system around it: data, instructions, retrieval, tools, evaluation, security, and accountability.

AI explainer

What Are Golden Sets?

A golden set is a small, carefully curated, human-validated evaluation dataset used as the trusted answer key for an AI system. It is the “we know what right looks like” set.

Think of it like this: here are the examples we absolutely know the correct behavior for. If the AI cannot do well on this, we do not trust it yet.

What belongs in a golden set?

• Input: prompt, question, log, alert, document, image, ticket, or API request.
• Expected output: ideal answer, label, classification, score, or action.
• Metadata: category, difficulty, source, risk level, version, and date.
• Human judgment: review by people who understand the domain.
• Pass/fail criteria: what counts as correct, incomplete, unsafe, hallucinated, or blocked.

For AI security

I would split golden sets into several lanes: prompt injection, benign security Q&A, malicious cyber intent, ambiguous intent, false positives, and enterprise policy behavior.

The golden set becomes the regression test suite. Every time the prompt, model, RAG pipeline, classifier, policy, or tool access changes, the golden set tells you whether the system got better, worse, or quietly broke something important.

AI explainer

Standard Model vs Custom-Tuned Model

A standard model is a general-purpose model trained to handle a broad range of tasks. It is useful out of the box, but it does not automatically understand your company, your tone, your workflows, your risk tolerance, or your exact definitions of good and bad behavior.

A custom-tuned model has been adapted for a narrower purpose. That might mean it speaks in your preferred format, recognizes your categories, follows your escalation rules, or performs better on a specific domain such as AI security, SecOps triage, code review, policy mapping, or prompt-injection detection.

The honest answer is that tuning is not always the first move. Sometimes a strong base model plus good system prompts, RAG, tool design, and golden-set evaluation is better than rushing into fine-tuning with messy data.

AI explainer

What Is Used in Tuning a Model?

Tuning uses examples. Not vibes. Not wishes. Examples. The model needs to see the kind of input it will receive and the kind of output you want it to produce.

Common tuning materials

• Instruction/response pairs: “when asked this, answer like that.”
• Classification examples: inputs labeled as benign, malicious, prompt injection, policy violation, false positive, escalation required, and so on.
• Preference data: two or more responses ranked by humans so the model can learn what “better” means.
• Domain examples: security alerts, policies, tickets, code snippets, architecture notes, incident summaries, or red-team prompts.
• Negative examples: what not to do, including unsafe answers, hallucinated claims, wrong severity, or overblocking.
• Evaluation sets: held-out data used to test the model, not train it.

The part people mess up

You should not train on your final exam. Keep your golden set separate. If the model sees the exact answers during training, the score becomes theater. It looks smart because you leaked the test.

For security work, the data should include boring normal cases too. A model that screams “malicious” at every security question is not safe. It is just useless with confidence.

AI explainer

GGUF, Safetensors, RAG, and the Acronym Swamp

The AI world is full of file formats, serving layers, tuning methods, and retrieval tricks. The names sound worse than they are. Here is the useful map.

For a practical system, these pieces work together: documents become embeddings, embeddings go into a vector store, RAG retrieves the right chunks, the prompt tells the model what to do with them, and the golden set tells you whether the whole contraption actually works.

Product

MarkForge

MarkForge is a cross-platform desktop Markdown editor with first-class Mermaid diagram support — Monaco editing, live preview, Mermaid Studio, workspaces, export to HTML/PDF/DOCX/slides, and a command palette for every action.

Windows and Linux builds are available now; macOS is on the way. The product site hosts downloads, screenshots, and the in-app update manifest.

MarkForge on rickcollette.org · markforge.rickcollette.org · github.com/rickcollette/markforge

Infrastructure

Operator Tools

Good infrastructure tools remove repeat pain without creating a giant new platform to babysit. The best ones are boring in the right places: clear inputs, predictable output, useful errors, and no mystery state.

The common rule is simple: build the thing, make it understandable, make it useful, and leave the door open for weirdness.

splatplug, tfvenv, estuge, and chperm live in this lane.

Security

Encrypted Storage

Storage is not neutral. If sensitive data is central to the system, protection belongs in the storage concept instead of being treated as an optional wrapper added at the end.

The KayveeDB work is about making encryption part of the object model: memory, rest, server surface, and command-line workflows.

Games

Game Systems

Game infrastructure is still infrastructure. Turns, state, persistence, player actions, and simulation rules need clean contracts or the world becomes impossible to reason about.

drokkit, drokkitexamples, and kyngdum are the current public examples.

Future Tech

Thesis

Agentic AI and governed tool use become the control plane. Physical AI makes automation visible. Programmable biology and new compute widen the blast radius.

The readiness score below is a heuristic discussion model, not a prediction engine. It weights growth, capital, infrastructure pull, governance pull, and platform breadth.

Future Tech

Methodology

This explains how the evidence and readiness model are framed.