litespeed/update-object-cache-settings changes the object cache configuration: backend, host, port, credentials, database, lifetime, and the global and non-persistent group lists.
Call litespeed/test-object-cache-connection afterwards. Always.
At a glance
| Ability | litespeed/update-object-cache-settings |
| Toolset | LiteSpeed Cache — toolset/litespeed |
| Group | LiteSpeed Cache › Object & Browser Cache |
| Requires | LiteSpeed Cache, active on the site |
| Capability | manage_options |
| Safety | Writes data |
| Repeatable | Idempotent — running it twice leaves the same result |
| Ships in | AcrossAI Abilities Manager 0.0.34 and later |
A wrong host is silent
There is no failure mode here that announces itself. Point the object cache at a host that does not answer and WordPress falls back to the database; the site keeps working and gets slower, and the configuration screen shows exactly what you typed.
So the test call is not optional diligence, it is the second half of the operation.
The group lists are the subtler settings. Non-persistent groups are the ones that must not survive the request — anything holding per-request state — and putting the wrong group there costs performance, while leaving a genuinely per-request group out of it can serve one request’s data to another.
Global groups matter only on multisite, where they mark the caches shared across the network rather than per site. Getting that wrong on a network is how one site reads another’s cached values.
The write is a patch, not a replacement. Send one key and one key changes; everything else in the area is left alone. The response names the keys that actually moved, so a payload that matched the existing values comes back with an empty list rather than a misleading success.
Input
| Name | Type | Required | What it is |
|---|---|---|---|
settings | object | Yes | A map of setting key to new value. Keys outside this area are refused, and the refusal lists what the area accepts. |
What comes back
| Name | Type | What it is |
|---|---|---|
updated | array | The keys that actually changed. An empty array means every value you sent was already set. |
settings | array | Every setting in the area after the write, in the read shape. |
success | boolean | Whether the call completed. |
message | string | A one-line summary of what happened, suitable for showing a human. |
error_code | string | Present only on failure — a stable machine-readable reason, such as a missing field, a wrong field type or an unmet confirmation. |
Example
Ask your assistant:
Point the object cache at our new Redis host.
It resolves to one call:
{
"ability": "litespeed/update-object-cache-settings",
"parameters": {
"settings": { "object-host": "127.0.0.1", "object-port": 6379 }
}
}Worth knowing
- Always test the connection afterwards. Failure is silent.
- Non-persistent groups hold per-request state. Getting the list wrong can leak data between requests.
- Global groups matter on multisite.
- A patch — unsent keys are untouched.
Related abilities
- Test Object Cache Connection — the required second step
- Get Object Cache Status — read the current configuration
- Flush The Object Cache — clear stale entries after a change
Get the plugins:
AcrossAI Abilities Manager · AcrossAI MCP Manager
Browse the rest: Every ability, by category