§ How to vibe

browse: web playground/notebook: latitude storage: github > feed to mcp-playbook key: vault ~ dev

planner consultant qa/data/sec

1. llm as a generator vector comparison // similarity search

2. dev cost --> 0

dev --> planner how to be a good planner?

1. how to work with llm effectively problem understanding problem solution planning

??.
problem understanding goal: brief ------- -----> system overview (modeling/arch) ?? context understanding Qs/steps de xac dinh van de? from given facts --> --> problem

manual see the thing read docs try app look at db schema

docs ~ deepwiki chat with docs ask qs > questions to ask? > ??

feed facts, artifacts, ... to LLM chat with llm > questions to ask?

1. new sdlc workflow why new workflow? (indie) effectively/efficient way(example) to code with llm (team) chuan hoa documentation/workflow persist project artifact/tribal knowledge => build up project context

how (context) brief system design feature design+modeling (prd) + (test)

(impl) (adr; changelog) (test report)

1. setup+configuration config ide rule config project rule

Here's our findouts using LLM in programming.

steps

choose an agent & model config ide rule config mcp config project rule config mcp for remote resources