Describe the bug

On four different TP-Link Tapo P410M smart plugs (used as generation meters for a balcony solar setup), the Energy module consistently fails to parse and gets disabled shortly after startup/reconnect, with a JSON decoding error:

Error querying <ip> for modules 'Time, AutoOff, DeviceModule, Energy, Matter, PowerProtection' after first update: no digit after minus sign: line 1 column 228 (char 227)
Error querying <ip> individually for module query 'get_emeter_data' after first update: no digit after minus sign: line 1 column 99 (char 98)
Error processing Energy for device <ip>, module will be unavailable: get_emeter_data for Energy (error_code=INTERNAL_QUERY_ERROR)

and, on subsequent polls:

Error processing Energy for device <ip>, module will be disabled: get_emeter_data for Energy (error_code=INTERNAL_QUERY_ERROR)

This affects the resulting Home Assistant current_consumption / today's-/month's-consumption sensors, which sit permanently at unknown/unavailable, even though the device is otherwise fully reachable and reports live power correctly in the official Tapo app (confirmed side-by-side: Tapo app showed ~372 W while HA's sensor.<device>_derzeitiger_verbrauch stayed on unknown).

A full homeassistant.reload_config_entry (fresh reconnect) does not resolve it — the error recurs on the very next Energy query.

Affected devices

All four are the same model, P410M, connected locally via KLAP (device_family: SMART.TAPOPLUG, encryption_type: KLAP, login_version: 2, https: false). Since the same error/column offset pattern appears across four physically distinct units at four different IPs, this looks like a firmware/protocol edge case rather than a single defective device.

A fifth device on the same network, a P110M, is unaffected — its Energy/consumption readings work correctly and continuously.

Environment

• Home Assistant Core: 2026.9.1 (Supervised/HAOS)
• tplink integration manifest requirement: python-kasa[speedups]==0.10.2 (confirmed this is also the current latest release on PyPI — no newer version available to test against)
• Devices: TP-Link Tapo P410M (energy-monitoring smart plug), x4, firmware/hw revision not yet captured (happy to pull via kasa CLI if useful)
• Comparison device: TP-Link Tapo P110M, unaffected

What I've checked

• This is likely related to (but distinct from) the P110M JSON_DECODE_FAIL_ERROR issue fixed for get_energy_usage in Tapo P110M: Error processing Energy for device, module will be unavailable: get_energy_usage for Energy  #1243 / PR Fallback to other module data on get_energy_usage errors #1245 (landed in 0.9.0) — that fix is present in our installed 0.10.2, but this P410M case still fails, with a different error code (INTERNAL_QUERY_ERROR) and a different JSON error (no digit after minus sign rather than a full decode failure), so it may be a different malformed-response shape not covered by the existing fallback.
• I don't yet have the raw pre-parse JSON payload from the device (only python-kasa's own exception message with a byte offset) — happy to capture a kasa --host <ip> --debug state (or equivalent) dump and attach it as a fixture if that would help reproduce/fix this, similar to the fixture used for Tapo P110M: Error processing Energy for device, module will be unavailable: get_energy_usage for Energy  #1243.

To reproduce

1. Set up a TP-Link Tapo P410M smart plug via the local tplink integration (KLAP auth).
2. Let it run for a normal polling cycle.
3. Observe the Energy module fail with the JSON decode error above and become permanently disabled until the plug is fully removed and re-added (a config-entry reload alone does not help).

Additional context

Happy to provide more diagnostics (raw device response capture, full debug log excerpt with homeassistant.components.tplink: debug / kasa: debug already enabled) on request.