Tmunzer Aide Mist Skills Mist Sle

Mist SLE Analyser

Unified SLE workflow for wireless, wired, and WAN: org-level site ranking, site drill-down, classifier breakdown, and impacted entity analysis.

SLE Domains

See references/metrics.md for full metric details, classifiers, and interpretation guides.

Score thresholds

Workflow

Step 0 — Resolve org_id and sites

1. get_mist_self → org_id.
2. If user names a site, resolve it by name: search_mist_data(scope='org', search_type='sites', org_id=..., filters={name:'<site_name>'}) → site_id. For the org-level SLE overview (Step 2), you'll need the full site list for the name map: get_mist_config(resource_type='sites', scope='org', org_id=..., limit=200). Paginate if has_more.

Step 1 — Determine the SLE domain

From the user's question, identify the domain:

• Wireless: mentions coverage, capacity, roaming, throughput, AP health, Wi-Fi SLE, wireless assurance
• Wired: mentions switch health, switch SLE, wired assurance, switch throughput
• WAN: mentions gateway health, WAN link, application health, WAN SLE, SD-WAN

Set sle_domain to "wifi", "wired", or "wan".

Step 2 — Org-level SLE overview

get_mist_insights(
  insight_type="sle",
  org_id=...,
  params={"query_type": "sites_sle", "sle": "<sle_domain>"}
)

Returns an array of site objects. Sites with no data return only {"site_id": "..."} — skip these.

Sites with data contain metric scores as decimal values (0.83 = 83%), plus num_aps / num_switches / num_gateways and num_clients.

Build a ranked scorecard sorted by lowest composite SLE. Composite = arithmetic mean of all numeric metric scores present for that site (e.g., if a site has coverage=0.83, capacity=0.75, throughput=0.99, composite = (0.83+0.75+0.99)/3 = 85.7%). Use site name map for display.

Step 3 — Select site(s) to drill into

• If user named a site → resolve its site_id.
• If not specified → drill into the worst-ranked site automatically.
• If multiple sites are red (< 80%) → drill into all of them.

Step 4 — Get enabled metrics for the site

get_mist_insights(
  insight_type="sle",
  site_id=...,
  params={"query_type": "metrics", "scope": "site", "scope_id": "<site_id>"}
)

Returns {"enabled": [...], "supported": [...]}. Only proceed with metrics in enabled that belong to the current domain. Prefer versioned metrics (e.g., switch-health-v2 over switch-health, wan-link-health-v2 over wan-link-health).

Step 5 — Per-metric summary and classifiers

For each enabled metric (skip if score >= 95% unless user asked for full breakdown):

get_mist_insights(
  insight_type="sle",
  site_id=...,
  duration="1d",
  params={
    "query_type": "summary",
    "scope": "site",
    "scope_id": "<site_id>",
    "metric": "<metric_name>"
  }
)

Extract:

• data.impact.num_users / total_users → % users/clients impacted
• data.impact.num_aps / total_aps → % entities impacted (APs, switches, or gateways depending on domain — the field name is always num_aps/total_aps regardless of domain)
• data.classifiers[] → each with name, impact.num_users, impact.num_aps

Compute classifier contribution: classifier.impact.num_users / metric.impact.total_users * 100.

Step 6 — Impacted entities (for metrics < 90%)

The query_type depends on the domain:

get_mist_insights(
  insight_type="sle",
  site_id=...,
  duration="1d",
  params={
    "query_type": "<entity_query>",
    "scope": "site",
    "scope_id": "<site_id>",
    "metric": "<metric_name>"
  }
)

Returns array: {mac|name, degraded, total} per entity. Compute per-entity SLE%: (1 - degraded/total) * 100. Rank worst-first, show top 5.

For "list APs/switches with SLE below X%": paginate fully and filter entities where computed SLE% < threshold.

Step 7 — Org-wide entity ranking (optional)

If user asks "list all APs in the org with coverage SLE < 40%":

1. Get the org scorecard (Step 2) to find all sites with data.
2. For each site, run Step 6 with impacted_aps and the requested metric.
3. Aggregate all entities across sites. Filter by threshold.
4. Present with site name for context.

WAN-specific notes

• wan-link-health-v2 = 0 often means "no WAN paths monitored" — not an outage. Cross-check with v1 and gateway-health.
• gateway-health = 0 usually means the gateway is offline.
• Use impacted_interfaces for per-interface breakdown on WAN link metrics.

Output

Use canvas (web-artifacts-builder) for multi-metric dashboards with score rings, classifier bars, and entity tables. The dashboard should use CSS variables for theming (see web-artifacts-builder skill). Use markdown for simple single-metric answers.

After any visual dashboard, write a brief prose summary (2-4 sentences) covering only the critical issues and recommended actions. Don't repeat what the dashboard shows.

Error handling

Source: tmunzer-AIDE/mist-skills — distributed by TomeVault.