Writing
Engineering disciplines, agent architecture, and what actually works.
The 7 Levels of Agent Engineering
Why most people stop at L2 and think they're at L5.
Read →L1-L2: Prompts and Tools
The floor everyone thinks is the ceiling.
Read →L3: Context Engineering
Second brain, not second mind.
Read →L4: Harnesses
Runtime control over your agent's mind.
Read →L5: Admissibility
Why your agent hallucinates and how to stop it.
Read →L6: Concentration
How to make an AI surf its own mind.
Read →L7: Emergence
When the system starts building itself.
Read →The Fork
Why most people should stop at L4.
Read →SANCTOT: A Complete L5 Example
From 'dad always has snacks' to a working ontology compiler.
Read →The Fair Game
Six levels of economic participation, seven levels of complexity, and the punchline nobody in the AI agency world wants you to understand.
Read →Holographic Work
Every piece contains the whole. One conversation produces everything.
Read →The 7 Engineering Disciplines
Prompt, Context, Tool, Harness, Admissibility, Concentration, Emergence.
Read →The Single Biggest Trick with LLMs
How to grow a metalanguage with an AI without fighting it — and what to call the things you grow.
Read →The Single Biggest Trick with LLMs, Part 2
Who I was when I wrote Part 1, who I became reading Jani's scriptures, and the actual boot — an LLM driving three POSTs against a live Neo4j graph.
Read →The Single Biggest Trick with LLMs, Part 3
JaniScribe: a real Cybernet in a live Neo4j graph, booted and ticked once. The LLM is the external driver. The Cybernet is the entity. This post is the bijective bridge.
Read →SHELL
Turn any toolkit into a game your agent can play.
Read →Admissibility Engineering
If the hidden edge is not represented anywhere, retrieval cannot retrieve it.
Read →HIEL
AI randomness is energy, not noise. Harness stochasticity.
Read →HALO
The invisible seam between your context and the machine's context.
Read →Interaction Loop
Seven elements that turn fragile handoffs into bulletproof coordination.
Read →Progressive Disclosure Harness
AI agents get worse the more tools you give them. The fix: staged reveals.
Read →Composition
Working pieces don't make working wholes. Five requirements.
Read →SOSEEH
Four questions decompose any blob into solvable pieces.
Read →Allegorization Compiler
Forge frameworks from analogies across unrelated domains.
Read →DocMagic Stack
Where admissibility tools sit in the stack.
Read →OK Stable Signal
Stop building on hope. Four verification properties.
Read →Towering
You have a foundation problem, not an intelligence problem.
Read →Seam Repair
Seven-step protocol instead of a panic response.
Read →Flow
Five structural conditions. Meet all five and hours vanish into progress.
Read →Externalization
Your brain holds 4 things. Your systems hold everything.
Read →Calibration
Fitting clicks are right 75%. Desire clicks are right 20%.
Read →Meta-Cognitive Awareness
The prerequisite your entire operating system depends on.
Read →The Blanket
Membrane, not wall. The third option between starving and overwhelm.
Read →Reach
Your distraction is pre-conscious pattern assembly.
Read →Cognitive Thermal Dynamics
Your brain overheats from unprocessed complexity.
Read →Helming
Vision comes from completion. Finish the layer to see the next one.
Read →Crowning
The system starts maintaining you instead of the reverse.
Read →Mission Control
Three questions that kill drift. Every 10 minutes.
Read →The 30-Day System Audit
The right SOP to automate first is the most leaky one.
Read →