docs: close the no-remote gap and repoint the deleted phase-04 notes

`origin` is `git@git.reeseapps.com:services/dashboards.git` (`main`) and the history is
pushed, so gap 3 ("No remote yet") is gone and the remaining gaps renumber. The remote has
no CI, which is why the same paragraph now names `bash .agents/validate.sh` as the only gate.

The `/llm` layout note cited `NOTES-chart-set.md`, a pipeline stray that is deliberately not
tracked; it now cites the phase-04 task file where that decision was written.

LLM-Generated: true

AGENTS.md

@@ -269,7 +269,8 @@ per monitored host (systemd unit, own user, `--bind`/`--port`/`--token`/`--rotat
 `--status`/`--uninstall`; it stores the token in a root-owned 0600 file and hands the agent
 `--token-file`, never argv, and runs under `ProtectSystem=strict`).
 The quadlet unit + env template + nginx/SNI live in the deployment repo, like every other
-service here.
+service here. The source itself is `git@git.reeseapps.com:services/dashboards.git` (`origin`,
+branch `main`); there is no CI on it, so `bash .agents/validate.sh` is the only gate.
 
 ## Testing policy
 
@@ -426,7 +427,8 @@ them at the defaults. `/llm`'s hero `.sub` prints the rate cell's own `window` t
 always say whether it was scraped, measured `Δcounter/Δt`, or estimated (`llm_monitor`'s
 `setThroughput` idiom; dropping the tag to look cleaner is a regression, not a simplification).
 
-**The `/llm` page reads per engine** (phase 04, `NOTES-chart-set.md`): a row of two
+**The `/llm` page reads per engine** (phase 04, decided in
+`.agents/phases/complete/04_llm_chart_readability/01_chart_set_decision.md`): a row of two
 `Dash.hero()` throughput numbers (decode, prompt — deliberately **uncoloured**, because 0 tok/s
 on an idle server is healthy and `severity.py` has no throughput row), then a compact `.cells`
 list — label left, tabular value right, a `sparkline()` fixed at 160px so rows line up, and a
@@ -508,9 +510,7 @@ memory plot; the `cpu.usage_pct@coreN` series is still stored and still queryabl
    fleet (R9700 + Strix Halo iGPU; gateway has none). An NVIDIA host would report no GPU at
    all — needs a `nvidia-smi`/NVML or DCGM path, and a fixture from a real card before the
    field map is written.
-3. **No remote yet.** No `origin` is configured (`git remote -v` is empty) — the user is
-   providing it, so every phase so far is local-only history.
-4. **UPS history depth.** Events and the numeric series are stored; long-window battery
+3. **UPS history depth.** Events and the numeric series are stored; long-window battery
    health (capacity fade over months) has the rollups but no dedicated view.
 
 ## Locked decisions (and why)