aroya-reporting

A Claude Code / Cowork plugin that bundles the AROYA MCP server and a full set of facility-reporting skills so any AROYA user can produce beautiful, color-enriched HTML reports against the facilities their bearer token is scoped to.

What's in the box

MCP server (mcp-server/) — exposes your AROYA account through a local MCP. Tools land under the mcp__aroya__* namespace so the skills pick them up automatically. Key tools include:

• mcp__aroya__list_facilities, mcp__aroya__list_rooms, mcp__aroya__resolve_names
• mcp__aroya__get_facility_overview, mcp__aroya__get_all_facilities_overview
• mcp__aroya__get_environmental_summary, mcp__aroya__get_current_readings
• mcp__aroya__get_raw_data, mcp__aroya__get_trend, mcp__aroya__get_room_history
• mcp__aroya__compare_rooms, mcp__aroya__compare_zones, mcp__aroya__analyze_drip_drain
• mcp__aroya__search_issues, mcp__aroya__generate_facility_report, plus caching / rate-limiting primitives

Skills (skills/) — five skills that cover the full reporting stack:

Skill	What it does
aroya-cultivation-analysis	Phase-specific crop-steering analysis against AROYA best-practice targets
aroya-hg-deepdive	Multi-run comparative harvest-group analysis with scoring and rich HTML/Docx dashboards
aroya-device-health	Mesh telemetry audit (battery, RSSI, travel-time, data gaps) with a room-grouped dashboard
aroya-design-system	The AROYA visual token palette, chart styles, and HTML/Print templates every deliverable inherits
aroya-report-publish	Wraps any analysis in the canonical AROYA light+dark bilingual (EN/ES) layout and optionally publishes to GitHub Pages

Install

1. Drop the plugin into Claude Code or Cowork. Either install the .plugin bundle directly, or symlink/copy this directory into your plugin tree. Claude discovers the MCP via .mcp.json and the skills via skills/*/SKILL.md automatically.

2. Set your AROYA token. Pick one:

   Option A — env var (recommended):

   export AROYA_BEARER_TOKEN='...your token...'

   Option B — .env file:

   cp mcp-server/.env.example mcp-server/.env
   # then edit mcp-server/.env and replace your-token-here

3. Restart Claude. On first invocation of any mcp__aroya__* tool, the launcher creates a local virtualenv at mcp-server/.venv and installs mcp, httpx, and python-dotenv. Subsequent launches reuse the venv.

   Python requirement: Python 3.9 or newer on PATH (as python3 or python).

Try it

After install, try any of these phrasings in Claude — they trigger the bundled skills:

• "Generate a facility report for "
• "Run a device health audit on "
• "Compare my last 6 harvest groups"
• "Ship a bilingual report for " → invokes aroya-report-publish

Optional: publish reports to GitHub Pages

aroya-report-publish can push the light + dark HTML pair live to a GitHub Pages repo you own. Configure:

export AROYA_REPORTS_REPO=<your-gh-user>/<repo>   # required to enable publish
export AROYA_REPORTS_BRANCH=main                   # optional, default: main
export AROYA_REPORTS_PAGES_URL=https://custom.example.com   # optional override

You'll need the gh CLI authenticated (gh auth login). If AROYA_REPORTS_REPO is unset, the skill still builds the HTML files locally — the publish step simply no-ops with a hint.

Token scope

Your bearer token is yours. Whatever facilities it's authorized against are the only facilities the MCP will ever return — the plugin doesn't bundle any credentials of its own. Admin tokens see everything; scoped tokens see only their own tenants.

Troubleshooting

• "AROYA_BEARER_TOKEN is not set" — you haven't exported the env var or created mcp-server/.env. See step 2 above.
• First call after install is slow — that's the one-time venv bootstrap. Subsequent calls are instant.
• mcp__aroya__* tools don't appear in Claude — confirm .mcp.json sits at the plugin root and that Claude was restarted after install. Check Claude's MCP server logs for run.sh output.
• Reports look like stock matplotlib — the skills pull styles from aroya-design-system. Make sure that skill is present in skills/ and that your analysis script is importing from aroya-design-system:scripts/aroya_chart_style.py rather than using matplotlib defaults.

File layout

aroya-reporting/
├── .claude-plugin/
│   └── plugin.json             # plugin metadata
├── .mcp.json                   # MCP server registration
├── mcp-server/
│   ├── run.sh                  # bootstraps venv, execs the server
│   ├── requirements.txt
│   ├── .env.example
│   └── src/
│       └── aroya_mcp_server_extended.py
├── skills/
│   ├── aroya-cultivation-analysis/
│   ├── aroya-hg-deepdive/
│   ├── aroya-device-health/
│   ├── aroya-design-system/
│   └── aroya-report-publish/
└── README.md (this file)

License

Provided as-is for AROYA agricultural consulting use.