Drive the "Ask Your Docs" engagement end to end — the staged flow that turns the user's documents into a live, grounded, shareable bot, with the agent leading and a living ~/work/PLAN.md. Use at the start of a doc-qa session and to decide the next step at any point. Lead the user through it; don't wait to be told each move.
Scanned 9/5/2026
Install to Claude Code
npx -y skills add Podbay-Cloud/podbay --skill ask-your-docs-onboarding --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Ask Your Docs Onboarding?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/podbay-cloud-ask-your-docs-onboarding)More formats (shields.io, HTML) on the badges page.
---
name: ask-your-docs-onboarding
description: >-
Drive the "Ask Your Docs" engagement end to end — the staged flow that turns the
user's documents into a live, grounded, shareable bot, with the agent leading and a
living ~/work/PLAN.md. Use at the start of a doc-qa session and to decide the next
step at any point. Lead the user through it; don't wait to be told each move.
---
# Ask Your Docs — the guided engagement
You lead. The user has an outcome ("a bot people can ask about my stuff"), not a spec.
Drive a staged flow, check in at real decisions, and keep `~/work/PLAN.md` current so a
fresh session resumes oriented (which stage, which docs, which choices).
## Stages (track each in PLAN.md: todo / doing / done)
1. **Live** — dev server up, public URL from `podbay preview` in hand, greeting shown.
2. **Key** — confirm `$ANTHROPIC_API_KEY` is set (else the bot errors on answer; guide
them to add it + restart `pnpm dev`).
3. **Understand** — what docs? who asks? public or internal? one sentence each.
4. **Ingest** — get their docs in; report what indexed; offer to clear the demo docs
(`doc-ingestion` skill).
5. **Verify grounding** — ask 2–3 real questions TOGETHER; confirm correct + cited + a
clean "I don't know" on an out-of-docs question (`ground-and-cite`, `rag-quality`).
Do NOT declare success on "a chat exists" — success is a grounded, cited answer.
6. **Make it theirs** — persona/tone, scope, title, suggested questions, branding.
7. **Publish + hand off** — share the public URL; tell them who can now ask.
8. **Next** — offer the menu below; let them pick.
## The "where next" menu (offer concretely, 1 line each)
- Add more docs / organize by topic.
- Tighten refusals so it never guesses.
- Branding: name, colors, welcome message, suggested questions.
- **See UNANSWERED questions** → exactly what docs to add next.
- Embed the bot on a website (bigger lift — flag it).
- Capture leads from conversations (bridges to a CRM flow).
- Digest of questions to Slack/email.
## How to lead
- One short greeting, then act — the app already works.
- Check in on REAL choices (persona, scope, what's public) — don't assume; don't
bury the user in options either. Recommend, then confirm.
- Keep momentum: after each stage, state what you did + the next step, and update PLAN.md.
- The bot is the deliverable — always be moving it toward "live, grounded, shared."
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!