Add Datadog CLI-usage dashboard + JSON validity gate

aai_cli/output.py

@@ -205,30 +205,20 @@ def emit_error(err: CLIError, *, json_mode: bool) -> None:
             error_console.print(f"[aai.muted]Suggestion:[/aai.muted] {escape(err.suggestion)}")
 
 
-# The `assembly` wordmark for the bare-command welcome screen: a compact two-row
-# half-block letterform, tinted to the brand accent (see theme.BRAND). Interior
-# spaces are load-bearing (they separate the glyphs); trailing spaces are not, so
-# they're dropped to survive whitespace-trimming tooling.
-_BANNER = """\
-▄▀█ █▀ █▀ █▀▀ █▀▄▀█ █▄▄ █   █▄█
-█▀█ ▄█ ▄█ ██▄ █ ▀ █ █▄█ █▄▄  █"""
-
 # A one-line header: emoji + product + version, then the product tagline.
 _TAGLINE = "AssemblyAI from your terminal"
 
 
 def print_banner() -> None:
-    """Print the welcome header — a version + tagline line, then the `assembly` wordmark in
-    the brand accent (the bare-command welcome screen)."""
+    """Print the welcome header — a single emoji + product + version + tagline line
+    in the brand accent (the bare-command welcome screen)."""
     # highlight=False so Rich's repr-highlighter doesn't recolor the version digits or
     # the quoted tagline — the line stays a single muted tone behind the brand label.
     console.print(
         f"[aai.brand]🎙️  AssemblyAI CLI[/aai.brand] "
         f"[aai.muted]{__version__} — {_TAGLINE}[/aai.muted]",
         highlight=False,  # pragma: no mutate (purely cosmetic: toggles Rich repr coloring, not text)
     )
-    console.print()
-    console.print(Text(_BANNER, style="aai.brand"))
 
 
 def print_code(code: str, *, language: str = "python") -> None:

docs/datadog/README.md

@@ -0,0 +1,41 @@
+# Datadog dashboard — CLI usage & reliability
+
+`cli-usage-dashboard.json` is an importable Datadog dashboard for the `aai` CLI's
+anonymous usage telemetry (`aai_cli/telemetry.py`). The CLI ships one log per
+command run to the Datadog **Logs** intake (`source:aai-cli`) — it is Logs, **not**
+RUM. Each log line = one command run.
+
+## Prerequisite: create the facets
+
+Log-analytics widgets only work on attributes that are **facets**. Before the
+dashboard shows data, add these in **Logs → Configuration → Facets** (or expand any
+`source:aai-cli` log and click each attribute → *Create facet*):
+
+- Facets: `@command`, `@outcome`, `@cli_version`, `@os`, `@python_version`,
+  `@exit_code`, `@ci`, `@device_id`
+- Measure (numeric, not a plain facet): `@duration_ms`
+
+`@device_id` (unique-device counts) and `@duration_ms` (latency percentiles) are the
+two that matter most.
+
+## Import
+
+UI: **Dashboards → New Dashboard → ⚙️ → Import dashboard JSON**, paste the file.
+
+API:
+
+```sh
+curl -X POST "https://api.datadoghq.com/api/v1/dashboard" \
+  -H "DD-API-KEY: $DD_API_KEY" -H "DD-APPLICATION-KEY: $DD_APP_KEY" \
+  -H "Content-Type: application/json" \
+  -d @docs/datadog/cli-usage-dashboard.json
+```
+
+## Notes
+
+- "Error rate" counts any `@outcome != success` — that includes expected user errors
+  (e.g. `auth_failure`), not just crashes. Filter to `@outcome:internal_error` for a
+  crash-only view.
+- `@device_id` is a per-machine random UUID, so it approximates installs, not people.
+- Events take ~1–2 minutes to index; a fresh `source:aai-cli` search can look empty
+  for a minute even when delivery succeeded.