The work moment
You are drafting a progress note or donor brief. It compares electricity access across five programme countries. Every figure has to trace back to the UN SDG Global Database.
The stored source gives each value with its Nature flag. Two examples for Kenya:
Kenya, 2015: 41.6 percent. Proportion of population with access to electricity, by urban/rural (%). Nature: Country data. Source: World Bank.
Kenya, 2024: 77.0 percent. Proportion of population with access to electricity, by urban/rural (%). Nature: Modeled data. Source: World Bank.
What the test showed
Claude ran with the skill and two tools allowed: WebFetch and Bash(curl:*). It was asked on 2026-10-30 at 18:26 UTC. It opened with this:
I couldn't create the Excel file. Running Python and
zipboth need your approval in this session, and neither was approved. I did download the data from the UN SDG Global Database and wrote the script that builds the workbook.
Then it gave the table. These rows are quoted word for word:
| Country | 2015 value (%) | Latest year | Latest value (%) | Change (pp) |
|---|---|---|---|---|
| Kenya | 41.6 | 2024 | 77.0 | +35.4 |
| Ethiopia | 29.0 | 2024 | 56.6 | +27.6 |
| Bangladesh | 74.0 | 2024 | 99.5 | +25.5 |
| Nepal | 82.0 | 2024 | 97.9 | +15.9 |
| Rwanda | 22.8 | 2024 | 72.0 | +49.2 |
2024 is the latest year with data for all five countries. The database lists the World Bank as the underlying source for every value.
Every 2015 and 2024 value in the table matches the stored source.
Where the answer stops
The Excel file was not created. Claude also said it could not run its usual checks:
Usually I'd also recalculate the formulas and run structural checks, but those tools (
recalc.pyandMiniMaxXlsx) aren't in this directory, so the file won't get those checks either way.
So approve Python, or run the script yourself, before the workbook goes anywhere.
Add the Nature flag yourself
The table has no Nature column. The stored source has one for every value.
- Kenya: Country data in 2015, Modeled data in 2024
- Ethiopia: Country data in 2015, Modeled data in 2024
- Bangladesh: Modeled data in both years
- Nepal: Modeled data in both years
- Rwanda: Country data in both years
If your brief compares the two years, say which values are modeled.
What Claude did with the skill
- Downloaded series EG_ACS_ELEC from the UN SDG Global Database, with Location = ALLAREA for the national total.
- Wrote a script that builds the workbook from the downloaded response.
- Planned a Cover sheet with the indicator, the source, the retrieval date and a colour key.
- Planned an Electricity Access sheet with the table. The change column is a live formula, latest value minus 2015 value.
- Kept the Provider and Reference Link columns.
- Offered to run the script once you approve Python, then reopen the file to confirm it worked.
The skill is minimax-xlsx by LeoYeAI/openclaw-master-skills, MIT licence.
Copy-paste prompt
Using the UN SDG Global Database, build one table in an Excel workbook showing the proportion of the total population with access to electricity (series EG_ACS_ELEC, national total) for Kenya, Ethiopia, Bangladesh, Nepal and Rwanda. For each country, give the 2015 value, the latest year with data, the value for that latest year, and the change in percentage points. Add Provider and Reference Link columns for the source.
Sources
- Proportion of population with access to electricity, by urban/rural (%)
- Proportion of population with access to electricity, by urban/rural (%)
- Proportion of population with access to electricity, by urban/rural (%)
- Proportion of population with access to electricity, by urban/rural (%)
- Proportion of population with access to electricity, by urban/rural (%)