Set Up the Repo

Configure a repository, once, so every other skill can read its answers instead of guessing. The output is a set of agent-facing config files under docs/agents/, seed spec files, and a pointer block in the repo's agent instructions.

This is a conversation, not a script: explore, present what you found, decide one thing at a time with the user, then write.

Template seeds live in this skill set's templates/ directory. Resolve pack seeds in this order, first path that exists: (1) templates/ beside this SKILL.md, (2) ${CLAUDE_PLUGIN_ROOT}/templates when that variable is set, (3) ../../../templates relative to this SKILL.md.

Track progress

This skill has seven steps (decisions A–L inside step 2) and skipping one is the common failure — an unconfigured tracker, or the Step 6 verification gate never run. Before Step 1, create a todo for each numbered step below and complete them in order, checking each off only when its Done when is met. Step 6 (prove the configuration works) is not optional.

1. Read the setup state

This step does one thing: determine whether the repo is already configured for the skill set, and how completely. Read the repo's own files only — do not probe external services or their auth (no gh auth, gh label list, glab, no Linear MCP call) and do not read the user's shell environment (no env, no *_API_KEY probing). Detection is not the job here: the user drives what gets set up in Step 2, and any service or toolchain specifics are gathered later, in service of a choice the user has already made.

Check the setup markers — all by reading files in the repo: