feat(mcp): diagnostic tools — overview, hosts, monitors, alerts, metrics, audit

MCP-4 of milestone 461 "MCP server: Steward as an LLM-inspectable,
LLM-operable hub". Twelve read-only viewer tools:

- get_overview: what is wrong now, ranked critical-then-oldest. Failing
  monitors with their error and when the run of failures began, firing,
  pending and still-breaching acknowledged alerts, TLS certificates expiring
  within 14 days, agents that stopped reporting, hosts over the configured
  CPU/memory/disk thresholds (Settings > Thresholds), failed Ansible runs in
  the last 24h, and findings from enabled plugins through a new optional
  `mcp_overview(db)` plugin hook. Each finding names the call that shows
  its evidence.
- list_hosts, get_host, get_host_metrics
- list_monitors, get_monitor (HTTP header values and URL passwords are
  redacted)
- list_alerts, get_alert_history, list_alert_rules, list_maintenance_windows
- query_metrics: the whole metric store, narrowed one argument at a time
  (sources > resources > latest values > a series summary)
- list_audit_events

Shared shaping in steward/mcp/tools/_shape.py: UTC ISO times with age_s,
bounded lists that say when and how they were cut, series as
min/avg/max/p95/last plus a ≤60-point series, lookup by id or
case-insensitive name with a did-you-mean, and a range argument that
refuses an unknown key instead of reading it as 24h.

New core reads in steward/core/metric_series.py (sources and resources
via loose index scans, latest values via DISTINCT ON, series across the
raw/hourly-rollup seam). monitors/service.py gains monitor_states,
down_since and recent_failures; agent_overview_by_host now carries the
fullest disk and the last sample time.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

docs/plugins/writing-a-plugin.md

@@ -356,6 +356,25 @@ async def list_targets(limit: int = 50) -> dict:
 - Refuse by raising `ValueError("what was wrong, and what to do instead")`. The agent reads that text. Any other exception reaches it only as a generic failure.
 - Tool names are global across core and every plugin; a name already taken fails registration and the plugin contributes no tools (logged).
 - Hot-reloading the plugin replaces its tools.
+- Follow the response conventions the core tools use (`steward/mcp/tools/_shape.py`): times via `iso()` plus `age_s()`, lists through `bounded()` so a cut is never silent, series through `summarize()`, and lookups by name or id through `pick()`.
+
+### Contributing to get_overview
+
+`get_overview` is the agent's first call: what is wrong right now, ranked. A plugin adds its own problems to it by exporting an optional `mcp_overview(db)` that returns findings built with `steward.mcp.tools.overview.finding`:
+
+```python
+# plugins/myplugin/__init__.py
+async def mcp_overview(db) -> list[dict]:
+    from steward.mcp.tools.overview import finding
+    return [
+        finding("warning", "target_unreachable", t.name, f"{t.name} has not answered for {t.misses} scans",
+                since=t.last_ok_at, ids={"target_id": t.id}, next=f'get_target(target="{t.id}")')
+        for t in await _unreachable_targets(db)
+    ]
+```
+
+- `severity` is `"critical"` (something is down or about to be) or `"warning"` (degraded, or heading that way). Report problems only; a healthy plugin returns `[]`.
+- It is called only while the plugin is enabled, with its own session. If it raises, the overview still returns, and names the plugin under `unavailable`.
 
 ---