profile-repolisted

# Resolve the repo profile (config for the issue/PR lifecycle skills) `create-issue`, `implement-issue`, and `merge-pr` are generic *workflows* wrapped around a thin layer of **repo-specific facts** — the commit author line, how to build and test, which label means "high priority", whether the repo squashes or rebases, which files conflict and how to resolve them. Hardcoding those welds the skills to one repo. This skill lifts them into a single committed file, **`.claude/skills/repo-profile.md`**, that the three skills read at their preconditions step. Drop the four skills into a new repo, run this once, and the lifecycle skills speak that repo's language. The profile is **data, not a skill** — plain markdown with no `SKILL.md`, so the loader ignores it; being committed, it travels with the repo. Because it exists, the lifecycle skills read it with a plain `cat` and only fall back to this skill when it's missing — so most of the time this skill isn't even invoked. ## Do this A bundled script does the deterministic work so you spend tokens on judgement, not probing. Run it from anywhere in the repo (it anchors to the git root). `<skill-dir>` is this skill's base directory — given when the skill loads: ```bash bash "<skill-dir>/scripts/repo-profile.sh" show ``` - **It printed the profile** (and no `--refresh` was asked) → **you're done.** Relay the headline values (repo slug, commit identity, build/test/format commands, integration style) so the caller sees what's in