Click Run blog
Field notes for local AI apps and diagnostic wrappers.
Practical posts for Click Run products: setup, troubleshooting, workflow estates, dependency maps, and local-first operations.
Click LLM
Local open-weight LLM picks by hardware tier
A maintained Click LLM guide for choosing Ollama models by RAM, disk, GPU, and local workload.
Click Agent
Turning a hot agent tool into a Hood2 execution graph
Popular agent tools often hide repeatable procedure inside prompts and tool calls. A Hood2 graph makes the deterministic part visible, testable, and reusable.
Diag n8n
Turning official n8n tutorials into a Hood2 dependency map
Official n8n tutorials are useful learning paths. Diag n8n turns a few of them into hierarchy nodes with explicit runtime, API, and credential dependencies.
Diag n8n
When n8n grows past one screen, put the workflow estate in a hierarchy
n8n is great for editing workflows, but a team with many workflows needs a second map: ownership, dependencies, prerequisites, and checks. Diag n8n uses Hood2 for that layer.
Click Claw
OpenClaw gateway not responding? The 4-layer check that finds it fast
When your OpenClaw gateway goes silent with no error, check it in dependency order — CLI, config, process, port — instead of guessing. Here's the exact sequence.
Click Claw
Why your local model (Ollama) hangs in OpenClaw — and how to fix it
Local LLM chats hang or spit out raw JSON in OpenClaw? It's almost never OpenClaw — it's the model. The three real causes (tool-calling, context window, size) and how to fix each.