onboard-our-stack
Onboard a newcomer to your codebase or answer a targeted 'how does our X work' question by walking curated reading lists — and cold-starting from the repo's own docs when no curated config exists yet.
Install
$ curl -fsSL https://raw.githubusercontent.com/Osipchuk/agent-skills/main/install.sh | sh -s -- onboard-our-stackOne line, works with any agent — installs just this skill into ~/.claude/skills/. Needs only curl; it fetches uv and the askill CLI if they're missing.
$ /plugin marketplace add Osipchuk/agent-skills
$ /plugin install skills@askillTwo slash commands inside Claude Code — installs the whole library as one plugin and lets Claude pick the right skill by context. No permission prompts, no remote-code-execution dance.
Installs skills/onboard-our-stack/SKILL.md into
~/.claude/skills/onboard-our-stack/; Claude reads it on the next
session start. The plugin tab pulls the whole library instead.
When it fires
Fires when someone asks how the team's own system works — broad ("I'm new, walk me through this project", "where do I start") or targeted ("how does our scheduler work"). With a committed .claude/onboard/topics.yaml it walks the curated path; without one it cold-starts from README/ARCHITECTURE/CLAUDE.md/docs and is explicit about the curated gotchas still missing. The newcomer creates nothing and changes no code.
name: onboard-our-stack
description: Onboard a newcomer to THIS codebase, or answer a targeted "how does our X work" question, by walking curated reading lists and synthesizing the mental model — never by guessing. Works out of the box: with a committed .claude/onboard/topics.yaml it walks that curated path; without one it cold-starts from the repo's own authoritative docs (README, CLAUDE.md / AGENTS.md, ARCHITECTURE.md, docs/, the dependency manifest, the entry point) and says plainly what curated knowledge is still missing. The newcomer never creates files, runs setup, or changes code. Use for "I'm new, help me understand this project", "where do I start", "onboard me", "explain our auth flow", "how does our scheduler work", "walk me through our X". Do NOT use for generic programming or public-library questions, debugging a specific error, or writing/modifying code. It teaches an existing system; it does not generate documentation or code, and it never invents gotchas or asserts a file is load-bearing beyond what the sources actually say.
What it does
- — Works out of the box: a newcomer just asks — no config to create, no setup, no code changes. The config is a maintainer's one-time committed artifact.
- — Cold-start fallback reads and CITES the repo's own docs rather than refusing; it never invents gotchas or guesses which files are load-bearing.
- — Broad mode walks an ordered curriculum (why -> getting-started -> one request end-to-end -> deep topics); targeted mode walks a single subsystem.
- — The keystone is request_lifecycle: one request narrated end-to-end across the layers — the story that actually hands over the mental model.
- — Curated knowledge (load-bearing files, ordering, the gotchas only in people's heads) stays human-authored and PR-reviewed, guarded by a stdlib CI check.