Skip to content
Agentmart
Submit a tool

onboard

First-run setup for the media-assistant-ops plugin.

Skill by danielrosehill

IndexedMIT
Write a review

Chalk Labs

Need help deploying onboard?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/danielrosehill-onboard

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

First-run setup for the media-assistant-ops plugin. Capture the user's local Music Assistant deployment URL + API key, snapshot the current speaker/player roster for future context, vet the overall setup (providers, connectivity, auth), and summarise the library (track/album/artist counts, active music providers). Use when the plugin is freshly installed, when `config.json` is missing, or when the user wants to re-baseline against a new server.

Use cases: backend, security

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/onboard

What's inside

Skill name
onboard
Description
First-run setup for the media-assistant-ops plugin. Capture the user's local Music Assistant deployment URL + API key, snapshot the current speaker/player roster for future context, vet the overall setup (providers, connectivity, auth), and summarise the library (track/album/artist counts, active music providers). Use when the plugin is freshly installed, when `config.json` is missing, or when the user wants to re-baseline against a new server.
License
MIT
Instructions
99 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Onboard — Music Assistant

End-to-end first-run: capture credentials, verify the server responds, snapshot the environment, and produce a context file the other skills can rely on.

## Phase 1: Capture credentials

1. Resolve the data root and ensure directories exist:

   ```bash
   DATA_ROOT="${CLAUDE_USER_DATA:-${XDG_DATA_HOME:-$HOME/.local/share}/claude-plugins}/media-assistant-ops"
   mkdir -p "$DATA_ROOT/data" "$DATA_ROOT/state"
   ```

2. If `$DATA_ROOT/config.json` exists, display the current values (redact API key to last 4 chars) and confirm the user wants to overwrite.

3. Ask the user for:
   - **Local deployment URL** — full base URL (e.g. `http://10.0.0.50:8095`). Accept an IP alone and assume `http://` + default port `8095`.
   - **WAN URL / hostname** (optional).
   - **API key** (required) — long-lived token from Music Assistant → Settings → Security.

4. Write `$DATA_ROOT/config.json` with `lan_url`, `wan_url`, `api_key`. `chmod 600` the file.

   Never echo the full API key back — only the last 4 characters for confirmation.

## Phase 2: Vet the setup

Run these checks against the LAN URL (fall back to WAN if LAN is unreachable and a WAN URL was provided). Report each as ✓ / ✗ with a short note.

1. **Reachability** — GET `/api/info` with the Bearer token. Capture server version, Python version, OS.
2. **Auth** — confirm the API key was accepted (200 vs 401/403).
3. **Provider health** — GET `/api/providers`. For each provider, report `name`, `type` (music / player / metadata / plugin), and `available` status. Flag any provider in an error state.
4. **Port / firewall** — if GET `/api/info` fails, attempt a raw TCP connect to the port and report whether the service is up vs whether it's an auth problem.

Produce a short **Setup Vetting Report** the user can read at a glance.

## Phase 3: Snapshot the speaker/player roster

1. GET `/api/players` and persist the full response to:

   ```
   $DATA_ROOT/data/players-snapshot.json
   ```

2. Write a human-readable summary to `$DATA_ROOT/data/players-snapshot.md` with one row per player:

   | Name | Player ID | Type | Provider | Default volume | Group membership |

   This file is the future-context artefact — other skills can grep it to resolve names → IDs and it gives you (Claude) a cheap way to know what hardware exists without a live call.

3. If any player is part of a sync group, note the group and which player is the leader.

## Phase 4: Library summary

1. Hit the library summary endpoints (if the running MA version exposes them) and persist raw JSON to `$DATA_ROOT/data/library-snapshot.json`:
   - GET `/api/library/tracks?limit=1` — read total count from pagination meta
   - GET `/api/library/albums?limit=1`
   - GET `/api/library/artists?limit=1`
   - GET `/api/library/playlists?limit=1`
   - GET `/api/library/radios?limit=1`

   If a given endpoint 404s on this MA version, skip it and note "not supported on this server version" in the report — don't fail the whole onboarding.

2. List active **music providers** (from Phase 2's provider fetch, filtered to `type=music`) with track count per provider if the API exposes it.

3. Write a human-readable `$DATA_ROOT/data/library-snapshot.md`:

   ```
   # Music Assistant Library — snapshot <ISO date>
   - Tracks: N
   - Albums: N
   - Artists: N
   - Playlists: N
   - Radios: N

   ## Music providers
   - spotify (connected) — 12,345 tracks
   - plex (connected) — 4,567 tracks
   - filesystem (connected) — 890 tracks
   ```

## Phase 5: Final report to the user

Print a single consolidated summary covering:

- ✓/✗ for each vetting check
- Server version + address used
- Player count and any notable states (offline, grouped, DSP-capable)
- Library totals and active music providers
- Paths of the snapshot files written under `$DATA_ROOT/data/`
- Next suggested commands: `/media-assistant-ops:status`, `/media-assistant-ops:list-players`

## Do not

- Do not write anything to `~/.claude/` or inside the plugin install directory.
- Do not echo the full API key after capture.
- Do not block onboarding if the library endpoints are unsupported — degrade gracefully with a note.
- Do not overwrite an existing `config.json` without explicit user confirmation.

From danielrosehill/Claude-Media-Assistant-Ops-Plugin, MIT.

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Apr 21, 2026 (6 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from danielrosehill

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get onboard running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying onboard?

CallGet help
Cite this page

onboard skill. Agent Mart. https://agentmart.dev/skills/danielrosehill-onboard (last updated 2026-10-24)