dashboardlisted
Install: claude install-skill zedarvates/botte-secrete
# Dashboard Skill
Live metrics visualization for botte-secrete — one data source, three views
(timestamped HTML report, ANSI TUI, live HTTP API). The live control room also
leads with a three-line Quality Compass: state, reason, and next safe action.
## Usage
```bash
python -m skills.dashboard.cli . # → .botte/reports/dashboard_<stamp>.html
python -m skills.dashboard.cli . --json # the assembled data
python -m skills.dashboard.cli . --tui # same data, rendered as ANSI panels once
python -m skills.dashboard.cli . --watch # re-render every --interval seconds (default 5s)
```
Assembles live numbers from `control_loop` (routing savings), `trends` (metrics
over time), `metrics` (LOC/cost), and `fix` (cost to apply outstanding fixes) into
one timestamped HTML page, browsable any time. Exposed via [[llm_mcp]] as
`dashboard`. Related: [[control_loop]], [[trends]], [[metrics]], [[fix]], [[report]],
[[demo]] (shares the same ANSI panel renderer), [[events]].
## `--tui` / `--watch`
`skills/dashboard/tui.py` renders the exact same `collect()` dict the HTML
report uses — one data source, two views. `--tui` prints one frame; `--watch`
loops, re-collecting and re-rendering every `--interval` seconds (Ctrl+C to
stop) — the metrics/routing/fixes counterpart to `demo --live`'s routing feed.
Metric panels include unicode sparklines (`▁▂▅▇`) sourced from `trends.show()`'s
last 10 snapshots — run `python -m skills.trends.cli snapshot .` periodically
(e.g. f