FIELD GUIDE / TASKS
Task hubs — phased findings, grouped by what you're about to do.
Each hub walks the lifecycle of a task and surfaces findings at the phase they bite. Empty cells are not hidden — they say where coverage is missing.
I'M ABOUT TO… 5 findings
Configuring agent autonomy
From default config to production controls — the gap between "agent stops at the boundary" and what really happens.
4 of 4 phases populated →
I'M ABOUT TO… 5 findings
Setting up MCP servers
From picking servers to running them in production — what breaks at each step.
4 of 4 phases populated →
I'M ABOUT TO… 5 findings
Building a RAG pipeline
From corpus to citation — the silent failure modes between an indexed document and a grounded answer.
4 of 4 phases populated →
I'M ABOUT TO… 4 findings
Picking an agent framework
From shortlist to production — where frameworks diverge and what breaks once an agent runs for real.
4 of 4 phases populated →
I'M ABOUT TO… 2 findings
Evaluating a benchmark
Before citing a public score — what the benchmark actually measures, and what it doesn't.
2 of 4 phases populated →
I'M ABOUT TO… 1 finding
Choosing an LLM gateway
From picking a gateway to running it for real workloads — what fails between the docs and production.
1 of 4 phases populated →