Skip to content

Query in Claude Code

New here? Instead of pinging the team, ask Claude Code. Nearly everything SCAILE knows lives in the scailetech GitHub org — this playbook, the company knowledge base, and a repo for every venture and client. Point Claude Code at the right repo, ask in plain language, and it answers from the source and cites where it found the answer.

The one principle

The answer is always in a GitHub repo. This playbook tells you how we work; the deeper "what" and "why" of a venture, a client, or the company itself lives in that thing's repo. So the move is always the same: find the right repo in the scailetech org, open it in Claude Code, and ask. When the playbook can't answer, it doesn't guess — it points you at the repo.

1. Query this playbook (fastest)

  1. Clone it (private, in the scailetech org — ask the current FA for access): git clone https://github.com/scailetech/playbook.git
  2. Open the folder in Claude Code.
  3. Ask in plain language — "how do I get access to the CRM", "who owns the GTM system", "what's the escalation path". Claude Code reads docs/, guided by the repo's root CLAUDE.md, and answers with a citation.

The repo's CLAUDE.md tells Claude Code to answer only from the playbook and, when it can't, to point you at the page Owner:, direct you to the relevant repo in the GitHub map, and log the gap in Improvements & ideas.

2. Query any other scailetech repo — the same move

The playbook is just one repo. The exact same "open in Claude Code, ask in plain language" pattern works against any repo in the org — and that's where you go the moment you need more than "how we work":

You want to know… Open / query this repo
SCAILE's canonical company context — GTM system, clients, team scaile-knowledge-base
How the site-builder engine works venture-factory
The GEO venture toolkit — Claude skills, GEO docs, article pipeline venture_skill_Anton
How a specific venture works — its brief, build, status that venture-* repo — see current ventures
A specific client engagement — scope, deliverables, data that client-* repo — see current clients

Don't know which repo? The GitHub map is the full tour of the org and how it's organized by naming convention (venture-*, client-*, gtm-*).

Clone and open — git clone https://github.com/scailetech/<repo>.git, open the folder in Claude Code, ask. Best when you'll actually work in the repo.

Reason across several at once — clone a few repos side by side under one parent folder and open that folder in Claude Code. Now it can answer questions that span them — e.g. the playbook plus scaile-knowledge-base plus the venture you're on — in a single ask.

3. Via MCP — query without cloning each time

Expose GitHub to Claude Code as an MCP source so any repo is queryable from any project, with no local clone:

  • A GitHub MCP server scoped to the scailetech org (or to a single repo like scailetech/playbook) lets you ask across the org's repos directly.
  • A filesystem MCP server pointed at a local clone's folder works if you'd rather keep files local.

Add the server to your Claude Code MCP config and ask from any project. There's no standardized org-wide MCP config yet — pick the GitHub or filesystem MCP that fits your setup, and if you settle on one worth standardizing, add it to the Improvements & ideas log. Setup details: Claude Code and MCP.

The rule, either way

Go to the GitHub repo. Ask the playbook first for "how we work". For anything deeper, the answer is in the source repo in the scailetech org — open it in Claude Code and ask the same way. If neither can answer, the page's Owner: line tells you who can, and the GitHub map tells you which repo should hold it. Either way, log what was missing in the ideas log so the next person doesn't hit the same wall.

Owner: Current FA Last reviewed: 2026-07-03