[github-actions]

A watched upstream document moved, or a fetch failure changed its
count, so upstream-docs-state.json no longer describes what the
watcher sees. Merging this makes the new observation the baseline the
next run compares against.

Read the upstream-doc issues this run opened before merging: this
accepts the new bytes as normal, and on its own says nothing about
whether the registers a driver decodes by have moved.

The required checks do not start on their own here. A pull request
opened with GITHUB_TOKEN starts no workflow run, so this one arrives
with an empty status rollup. Close and reopen it to run them. The
validate run dispatched against the branch head is a verdict you can
read in the meantime -- GitHub does not count it toward the rollup.

Opened by the watch-upstream-docs workflow.

[miravoss26]

Bot baseline update for `upstream-docs-state.json`. It refreshes sha256 and content length for three watched docs (Accuenergy Acuvim II manual, Easee observations, Easee ongoing session API) and adds six new watched URLs (HeishaMon MQTT topics, SunSpec model 802, SunSpec Device Information Model 1.1 and others). Only this state file changes.

• Low: the six new URLs point at raw.githubusercontent.com and sunspec.org. This file only records what the job fetched. The source list that adds them lives elsewhere, so it's worth confirming that list was reviewed when it changed.
• No CI checks on this head yet (mergeable is UNKNOWN).

Safe to merge from my read, as far as this file goes. device-drivers isn't on the autonomous merge allowlist, so a human should merge it.

[miravoss26]

Reviewed. This updates upstream-docs-state.json only — new/changed sha256 + content_length baselines for the watched upstream docs (heishamon, pixii, remne_p1ib, sungrow, tesla_vehicle additions; refreshed hashes for acuvim, easee, saj, teslamate). No driver code, no logic change.

• Correctness: matches the PR body's own description (accepting new observed bytes as the new baseline). One new entry (pixii::sunspec-device-info-pdf) has consecutive_failures: 1 / last_error: 403 Forbidden — that's an honest failure record, not a merge blocker, but worth a human glance if pixii driver work depends on that PDF.
• Security: no secrets, no code paths, no new runtime network calls — these are just tracked doc URLs for the watcher, all public, same shape as existing entries.
• Empty status rollup is expected per the PR body (GITHUB_TOKEN-opened PRs don't trigger workflows) — close/reopen to run checks if you want CI's own verdict before merging.

Safe to merge from my read.

[miravoss26]

Automated upstream-docs hash refresh (upstream-docs-state.json): content_length/sha256/last_changed bumps for several already-tracked driver docs, plus a handful of new entries (Easee SignalR stream, HeishaMon MQTT topics, Pixii SunSpec models, Remne P1IB readme/index, Sungrow manual, Tesla BLE proxy source). Pure data, no code.

• No secrets, no new dependencies, no new network destinations beyond already-tracked public doc URLs, no injection surface.
• One partial failure captured inline (Pixii SunSpec PDF → HTTP 403, consecutive_failures: 1) — that's the watcher correctly recording a fetch miss, not a bug introduced here.

Safe to merge from my read.

[miravoss26]

Bot baseline refresh for upstream-docs-state.json. Data only, no code.

• Real doc movement: Acuvim II manual (+1.5 MB), Easee observation IDs and ongoing-session API (the latter shrank 606 KB to 400 KB), and the NIBE myUplink changelog. Check the matching upstream-doc issues before merging, since this PR only accepts the new bytes and doesn't say whether register maps changed.
• New watched URLs: Easee x2, HeishaMon, SunSpec 802, Remne P1IB.
• One fetch failure to note: SunSpec Device Information Model PDF returns HTTP 403, so consecutive_failures: 1 and no hash.

Needs a human to read the upstream-doc issues first; not auto-mergeable.

[miravoss26]

Baseline update for upstream-docs-state.json (+193/-24). Data file only, no code.

• Changed: Accuenergy Acuvim II manual (18.6 to 20.1 MB) and the Easee charger observation IDs doc (272 to 320 KB) have new hashes. Three Easee doc URLs are newly tracked (api-and-streaming, signalr-stream, and one more).
• No upstream-doc issues are open in the repo, so nothing explains the Acuvim and Easee changes. Someone should diff those docs for register or observation-ID changes before accepting this as baseline.
• No checks ran on this PR (opened with GITHUB_TOKEN). Close and reopen to get them.

Needs a human decision: accepting it only makes the new bytes the baseline, it says nothing about whether the drivers are affected.

[miravoss26]

Correction to my review: I only saw the first part of the diff. The PR adds eight newly tracked URLs, not three: two Easee docs, HeishaMon MQTT-Topics, two Pixii/SunSpec docs, the Sungrow SHRT manual, and two TeslaBleHttpProxy source files. The rest of my read stands.