Document · doc-0003

Closed GitHub issues (pre-Backlog history index)

Every issue that existed on this repository’s GitHub tracker before it moved to Backlog.md on 2026-08-14. Kept as one document rather than as Done tasks on purpose: importing them as tasks would create a second ID space over the same history — backlog IDs follow creation order, so a mdh-NNNN could never be made to match the #NNN already cited in commit messages and code — and a wall of Done rows would drown out the board’s only real signal, which is what is left.

Where the bodies are

Mostly not on GitHub. The ten issues authored by the repository owner — #293 and #295–#303 — were deleted on 2026-08-14, immediately after this archive was captured, committed and pushed. gh issue view <N> 404s for those. The record is archive/github-issues-2026-08-14.json, which is redacted — Meraki identifiers, reporter handles and one reporter’s Home Assistant system-health dump were replaced with stable placeholders. See archive/README.md for the mapping and for what was deliberately left in.

Six issues were kept and still resolve on GitHub: #132, #143, #144 and #218 were filed by external contributors and are not ours to delete, and #50 and #256 are Renovate dependency dashboards, which are recreated on the next run regardless. They are in the archive too, so it is a superset.

jq '.[] | select(.number == 303)' archive/github-issues-2026-08-14.json
jq -r '.[] | select(.number == 303) | .comments[].body' archive/github-issues-2026-08-14.json

The tracker remains enabled for new external bug reports; it is no longer where this project’s own work is planned.

How to read the SHA column

No commit message in this repository cites an issue number. Every SHA below was matched by commit subject and date against the issue’s content and close date — it is an inference, not a recorded link. Treat a single SHA as “the change that appears to have closed this”, and verify before relying on it. Where nothing matched, the cell says so rather than guessing.

Note also that GitHub shares one number space between issues and pull requests, so the gaps in this table (#133–#142, #217, #264, #304–#308, …) are pull requests, not missing issues.

The index

# Title State Closed Resulting change (inferred)
50 Dependency Dashboard closed 2026-03-16 Renovate bot artifact; superseded by #256. Kept — still on GitHub
132 API Permissions (kept) closed 2025-10-12 No code change — answered: only org-level API keys are supported, network-level keys 404
143 MT20 Door Sensors (kept) closed 2025-10-17 c93df09 fix(sensors): correct MT device model capabilities mapping
144 MT20 is a door sensor (kept) closed 2025-10-18 c93df09, 9255697 fix(sensors): expand fallback capabilities
218 CW9171I not mapping (kept) closed 2025-12-20 8b0491a feat: add CW device support · 90e7bc2 feat: add device type mapping system · 3dd91c7 refactor: use centralized device type detection · 270c5bb test: add CW wireless device type tests
256 Dependency Dashboard open Renovate bot artifact. Kept — left in place and left open
293 Docs site: redesign & rebrand alignment + SEO/LLM discoverability closed 2026-07-03 e2d160c feat(docs): align docs site with m7kni.io brand + server-side SEO/LLM metadata · 8d54cc1 docs(geo): content-shape pass
295 MT-only major version (v1.0.0) — epic tracker closed 2026-07-03 a820159 feat!: MT-only major version — strip MR/MS/MV, overhaul MT API calling
296 Lane A — remove non-MT device files + de-wiring closed 2026-07-03 part of a820159
297 Lane B — const/types/config pruning closed 2026-07-03 part of a820159
298 Lane C — Meraki API overhaul (MT) closed 2026-07-03 part of a820159
299 Lane D — entities/factory + minimal-health sensors closed 2026-07-03 part of a820159
300 Lane E — config-entry migration 2→3 + repair closed 2026-07-03 part of a820159
301 Lane F — docs/branding + config_flow + diagnostics closed 2026-07-03 part of a820159
302 Add MT air-quality metrics: NO2, O3, PM10 closed 2026-07-03 b29c33e feat(mt): add NO2, O3, and PM10 air-quality sensors
303 MT sensors stop reading after 0.38.0: gateway-connections envelope wipes all readings closed 2026-07-03 ac76e8a fix(mt): survive gateway-connections {items,meta} envelope · db8d84e fix(mt): read gateway RSSI/last-seen from nested sensor.serial · b29dccd fix(mt): stop spamming ERROR when a sensor has no gateway

16 issues, 15 closed, 1 open. Ten deleted, six kept — rows marked (kept) still resolve on GitHub; every other row 404s and reads only from the archive. The epic #295 and its six lane issues #296–#301 all landed as the single squashed commit a820159; the lane split existed for the campaign, not in the history.

What this history is worth reading for

The MT-only narrowing (#295–#301) is the decision that still governs scope — it deleted MR/MS/MV support deliberately, and anything proposing to add another device family is reopening it. #303 is the most instructive bug in the repository’s history and is written up as a recurring defect class in the Wave operating model doc. #218, #143 and #144 are the model-and-metric mapping failures that made prefix matching a known-fragile surface.

View the source file on GitHub