Claude to reconcile Greek electricity bills
GitHub user pmoust used Claude Code to examine the electricity bills for four supplies in Greece: three with Heron and one with DEI. The problem was understanding the cost actually paid, including adjustments and conditions that a headline rate did not explain.
github.com
Original prompt
Shared by the source author. See source notes for context.
Read my bills in `./bills/*.pdf` and tell me, per supply, what I actually pay and whether I should switch.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- pmoust
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The owner reports a report and Excel comparison model for four actual Greek supplies; the public demo contains synthetic data.
- Human involvement
- Rate refreshes are proposed for human review; no completed provider switch is reported, and real bill outputs remain private.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Read each supply's PDF bills and reconcile its actual charge components.
- Model eligible alternatives with fixed charges, tiers and commitment terms.
- Review the report and proposed rate updates before making a provider decision.
GitHub user pmoust used Claude Code to examine the electricity bills for four supplies in Greece: three with Heron and one with DEI. The problem was understanding the cost actually paid, including adjustments and conditions that a headline rate did not explain.
The agent read PDF bills, reconciled the charges by supply, and produced a report and an Excel model for comparing alternatives. The comparison accounted for fixed charges, tariff tiers, commitment conditions and eligibility, rather than simply sorting advertised cents per kilowatt-hour. The repository describes the difference between an attractive advertised energy price and the realized energy cost on the author's bills.
The real bills and generated household outputs are private and excluded from the public repository. Its public demonstration uses synthetic provider data, so those screenshots should not be treated as verification of the author's private amounts. The firsthand README supplies the account of using the toolkit on the four actual supplies.
There is also a recurring update path: a smaller Claude model refreshes tariff data and proposes a pull request for human review. That is separate from automatically choosing or buying a tariff. The source reports bill reconciliation and comparison artifacts, without reporting that the author changed electricity provider.
Prompt
Original prompt
Source: github.com · pmoust
Frequently asked questions
Why was the headline rate insufficient?
The comparison included adjustments, fixed charges, tiers and commitment or eligibility conditions that affect the realized cost.
Are the public screenshots the author's bills?
No. The public demonstration is synthetic; the actual bills and household report are kept private.
Did an agent automatically change provider?
No switch is reported. The recurring rate updater proposes a pull request for human review.
More details
About this account
Claude parsed actual bills, reconciled supply costs and generated a comparison report and Excel model.
Firsthand source opened and read. Prompt is a verbatim excerpt of the published original instruction.
Credit: pmoust
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Python · pdftotext · Excel model
Electricity bills · Tariff comparison · PDF reconciliation
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.