oodle-dashboards

Oodle Dashboards — CRUD and Organization

This skill teaches the agent to manage Oodle dashboards and folders without losing panel state and without orphaning dashboards in the root folder.

Prerequisites

brew install oodle-ai/oodle/oodle
oodle configure
# or
export OODLE_API_KEY=<key>
export OODLE_INSTANCE=<instance>
export OODLE_DEPLOYMENT=<url>

Verify dashboards endpoint works:

oodle dashboards list -o json | jq 'length'
oodle folders list -o json | jq 'length'

Command Execution Order

Before running any oodle command:

1. Check whether the required resource ID or name is already in context.
2. If not, run the discovery command (e.g., oodle dashboards list -o json).
3. If the result is ambiguous, ask the user to confirm before proceeding.
4. Run the target command with the resolved ID.
5. Do not run speculative commands (e.g., do not delete without first get-ing the resource).

Quick Reference

Task	Command
List dashboards	oodle dashboards list -o json
Get dashboard	oodle dashboards get <uid> -o json
Create dashboard	oodle dashboards create -f save.json
Update dashboard	oodle dashboards create -f save.json with "overwrite": true (there is no dashboards update)
Delete dashboard	oodle dashboards delete <uid> --force
List folders	oodle folders list -o json
Create folder	oodle folders create --title "Platform" -o json
Delete folder	No CLI command; use the Oodle UI

Common Operations

Listing dashboards

# ✅ CORRECT
oodle dashboards list -o json

# ✅ CORRECT — filter by folder (the list also has folders, with `type` "dash-folder")
oodle dashboards list -o json | jq '.[] | select(.type=="dash-db" and .folderUid=="platform")'

# ❌ WRONG — parsing table output to find an ID
oodle dashboards list | grep "API Overview" | awk '{print $1}'

Reading a dashboard before changing it

# ✅ CORRECT — fetch the full definition first; preserves all panels and queries
oodle dashboards get dash_123 -o json > current.json
jq '{dashboard: .dashboard, folderUid: .meta.folderUid, overwrite: true}' current.json > save.json
$EDITOR save.json
oodle dashboards create -f save.json

# ❌ WRONG — sending an incomplete payload removes panels
oodle dashboards create -f <(echo '{"dashboard":{"uid":"dash_123","title":"new title"},"overwrite":true}')

Creating a dashboard

oodle dashboards create -f takes a save request: Grafana dashboard JSON in dashboard, plus folderUid and overwrite. A panel keeps its PromQL in targets[].expr. A complete save request places the dashboard in a known folder:

{
  "dashboard": {
    "title": "API Overview",
    "description": "Latency, error rate, and throughput for the API service.",
    "tags": ["service:api", "team:platform"],
    "panels": [
      {
        "title": "Request rate",
        "type": "timeseries",
        "targets": [{"refId": "A", "expr": "sum(rate(http_requests_total{service=\"api\"}[$__rate_interval]))"}]
      },
      {
        "title": "Error rate",
        "type": "timeseries",
        "targets": [{"refId": "A", "expr": "sum(rate(http_requests_total{service=\"api\",status=~\"5..\"}[$__rate_interval]))"}]
      },
      {
        "title": "P99 latency",
        "type": "timeseries",
        "targets": [{"refId": "A", "expr": "histogram_quantile(0.99, sum by (le) (rate(http_request_duration_seconds_bucket{service=\"api\"}[$__rate_interval])))"}]
      }
    ]
  },
  "folderUid": "platform",
  "overwrite": false
}

# ✅ CORRECT
oodle dashboards create -f save.json

# ❌ WRONG — no folderUid, dashboard ends up in root and is hard to find
oodle dashboards create -f <(echo '{"dashboard":{"title":"API Overview","panels":[...]}}')

Folder management

# ✅ CORRECT — create folder first, capture id, then create dashboards in it
FOLDER_UID=$(oodle folders create --title "Platform" -o json | jq -r '.uid')
jq --arg f "$FOLDER_UID" '.folderUid=$f' save.json > save.with-folder.json
oodle dashboards create -f save.with-folder.json

# ❌ WRONG — creating dashboards before folders, then trying to move them later
oodle dashboards create -f save.json
oodle folders create --title "Platform"

Safe deletion — two-phase

Dashboards are linked from runbooks, slack messages, and bookmarks. Delete in two phases.

# ✅ CORRECT — phase 1: rename so users see it's about to be removed
oodle dashboards get dash_123 -o json > dash.json
jq '{dashboard: (.dashboard | .title = "[MARKED FOR DELETION] " + .title), folderUid: .meta.folderUid, overwrite: true}' dash.json > dash.deletion.json
oodle dashboards create -f dash.deletion.json
# wait at least 7 days, then:
oodle dashboards delete dash_123 --force

# ❌ WRONG — immediate hard delete, breaks every existing link
oodle dashboards delete dash_123 --force

Best Practices

Never use $__range as a panel lookback window

$__range expands to the whole dashboard time range, so the window stays that wide at every step and each step re-reads almost the same data as the step before it. A 6h dashboard with a [$__range] window reads about 180 times more samples than the same panel with [$__dd_interval], and the multiple grows as the user zooms out — which is when the dashboard times out. Oodle marks such a panel with a warning triangle.

Use a window that scales with the time range instead:

Variable	What it is	Use it for
$__rate_interval	Grafana's built-in rate window, equal to max(step + scrape interval, 4 x scrape interval), so a rate always has enough samples to be defined.	rate() and increase() on a graph
$__dd_interval	The rollup window Oodle picks for the time range, matching the Datadog default. Capped at 4h.	The default rollup, and stat panel totals
$__large_interval	A coarser rollup than $__dd_interval for the same range. Capped at 12h.	Bar charts and long lookbacks

The two Oodle rollups resolve like this:

Dashboard range	$__dd_interval	$__large_interval
1 hour	20s	1m
6 hours	2m	5m
24 hours	5m	20m
7 days	1h	4h
30 days	4h	12h

All three grow and shrink with the dashboard time range, so the samples read per step stay about the same however far the user zooms out.

# ✅ CORRECT — the window shrinks and grows with the dashboard range
"expr": "sum(rate(http_requests_total[$__rate_interval]))"

# ✅ CORRECT — coarser rollup for a bar chart
"expr": "topk(10, sum by (route) (rate(http_request_duration_seconds_sum[$__large_interval])))"

# ❌ WRONG — one flat line at the range average, read 180 times over
"expr": "sum(rate(http_requests_total[$__range]))"

# ❌ WRONG — `$__range_s` is there to undo an `increase`; use `rate` instead
"expr": "sum(increase(http_requests_total[$__range])) / $__range_s"

Total a stat panel over windows, not over $__range

A stat panel that shows one number for the whole range does need every sample, but it does not need to read them in one step. Split the range into windows and add the windows back up.

# ✅ CORRECT — window == min step, range query, reduced back to one number
"expr": "sum(increase(http_requests_total[$__dd_interval]))"
"interval": "$__dd_interval"
"instant": false
"transformations": [{"id": "reduce", "options": {"reducers": ["sum"]}}]

# ❌ WRONG — instant query that reads the whole range on every refresh
"expr": "sum(increase(http_requests_total[$__range]))"
"instant": true

Keep the min step equal to the window, so the windows tile the range: a smaller step double counts samples, a larger one skips them. Match the reducer to the function — sum for increase, sum_over_time and count_over_time, max for max_over_time, min for min_over_time, mean for avg_over_time. Do not split quantile_over_time, changes or resets this way; the reduced value is not the value over the range.

Always get before update to preserve panel configuration

A save with "overwrite": true is a full-document replace. Missing panels in the payload will be removed.

# ✅ CORRECT
oodle dashboards get dash_123 -o json > dash.json
jq '{dashboard: (.dashboard | .panels[0].title = "Request rate (per second)"), folderUid: .meta.folderUid, overwrite: true}' dash.json > dash.new.json
oodle dashboards create -f dash.new.json

# ❌ WRONG — sends a single-field payload; all panels disappear
oodle dashboards create -f <(echo '{"dashboard":{"uid":"dash_123","description":"updated"},"overwrite":true}')

Always set folderUid when creating a dashboard

Dashboards in the root folder are hard for teams to discover.

# ✅ CORRECT
"folderUid": "platform"

# ❌ WRONG — omitting folderUid means root
"folderUid": null

Use the two-phase rename → wait → delete pattern for shared dashboards

Hard-deleting a dashboard breaks every external link (runbooks, slack reactions, bookmarks).

# ✅ CORRECT — phase 1: rename
oodle dashboards create -f dash.deletion.json
# wait, confirm no traffic, then phase 2: delete
oodle dashboards delete dash_123 --force

# ❌ WRONG — same-day delete on a shared dashboard
oodle dashboards delete dash_123 --force

Tag dashboards with service and team labels

Tags make dashboards searchable and let other tools (e.g. service catalogs) link to them.

# ✅ CORRECT
"tags": ["service:api", "team:platform", "env:prod"]

# ❌ WRONG
"tags": []

Failure Handling

Error	Cause	Fix
401 Unauthorized	Invalid or missing API key	Run oodle configure or set OODLE_API_KEY
404 Not Found	Dashboard or folder ID does not exist	Verify with oodle dashboards list -o json
connection refused	Wrong OODLE_DEPLOYMENT URL	Check OODLE_DEPLOYMENT env var
folder not found	folderUid references a deleted folder	List folders with oodle folders list -o json; choose an existing id or create one
Panels disappeared after update	A save with "overwrite": true had a partial payload	Re-create from the last get snapshot; in the future always get → edit → save
Cannot delete folder	Folder still contains dashboards, and the CLI has no folders delete	Move or delete the dashboards first (`oodle dashboards list -o json
429 Too Many Requests	Bulk dashboard sync	Add --retries 3, throttle to <10 creates per second

References

• Oodle CLI repo
• Oodle docs
• Query optimization — lookback windows and $__range