Sheet Tracker Creator

This skill turns the finished backlog into the artifact the PM actually opens every day: a shared tracker spreadsheet. It has two paths — a live Google Sheet when a Google MCP is connected, and a local .xlsx fallback when it is not — so it always produces something usable.

The tracker schema

Exactly four sheets, in this order:

Step 1: Assemble the tracker data

Collect the estimated backlog into one JSON object matching the schema the builder expects. The scripts/build_tracker_xlsx.py header documents the exact shape, and examples/ in the repo contains a filled example. In short:

{
  "project_overview": {"Project Name": "", "Client": "", "Objective": "", "Priority": ""},
  "epics": [{"epic_id": "EP-01", "epic_name": "", "description": ""}],
  "user_stories": [{"story_id": "US-01", "epic": "EP-01", "story": "", "priority": "P0", "story_points": 5}],
  "risks": [{"risk": "", "impact": "High", "mitigation": ""}]
}

Step 2: Choose a path

Check for a Google MCP first. If a Google Drive/Sheets MCP is connected, use Path A. Otherwise use Path B. Tell the user which path you took and why.

Path A — Live Google Sheet (preferred)

Follow this sequence with the connected MCP tools (exact tool names vary by server — check the server's tool list; see references/sheets-mcp-setup.md):

1. Create the spreadsheet in the designated Google Drive folder, titled [Project Name] — Project Tracker.
2. Create the four worksheets named exactly Project Overview, Epics, User Stories, Risks. Remove the default empty sheet.
3. Populate each worksheet: write the header row, then the data rows, using a batch update so it is one call per sheet, not one per cell.
4. Share the spreadsheet with the project team (writer access) or set link sharing per the client's data-access rules.
5. Return the URL. This is the headline deliverable — surface it prominently.

If any MCP call fails, do not silently drop the tracker: report the failure, then fall back to Path B so the user still gets a spreadsheet.

Path B — Local .xlsx fallback

When no Google MCP is available, generate the workbook locally:

python scripts/build_tracker_xlsx.py --input tracker_data.json --output project-tracker.xlsx

The script builds all four sheets with header formatting, sensible column widths, and frozen header rows, then reports the output path. Hand the file to the user and note that they can upload it to Google Drive and open it as a Sheet, which preserves the structure.

Step 3: Confirm and summarize

Report: the tracker location (URL or file path), the four sheet names, and row counts (number of epics, stories, risks). Confirm sharing was applied.

Example

For the pharmacy reorder project, the tracker ends up with a Project Overview naming ClientCo and the objective, an Epics sheet with EP-01..EP-05, a User Stories sheet listing each story with its P-level and points, and a Risks sheet capturing the bKash sandbox risk and the Ramadan deadline risk.

Troubleshooting

• MCP connected but calls fail. Verify the server shows "Connected" and that the account has Drive write scope. Test the MCP directly ("create a blank sheet") to isolate whether the issue is the MCP or this skill. Then fall back to Path B.
• Wrong sheet or column names. The schema is fixed — the four sheet names and their columns must match exactly so downstream tools and reviewers can rely on them.
• Data-access restrictions. If the client prohibits certain data leaving their environment, prefer the local .xlsx path and let the user place it themselves.