Set up the Redis object cache with Kimi Code

Setting up a Redis object cache with Kimi Code means asking Moonshot AI's terminal agent to check, test, enable or disable the object cache on a WordPress site through xSpeed Hub, from the session where you work on that site.

get_object_cache_statustest_object_cacheget_settingsupdate_settingstoggle_object_cacheflush_object_cache

Last updated: October 2026

Example session · Kimi Code

You: Use xSpeed Hub to turn the object cache off on staging before I clone it, and confirm the drop-in is gone.

get_object_cache_status(site: staging) read-only
→ Drop-in installed. Backend: Redis, connected.

Kimi Code: Disabling removes xSpeed's drop-in and its wp-config.php block on staging, and WordPress goes back to its own per-request cache. Disable the object cache on staging?

You: Yes, disable it.

toggle_object_cache(site: staging, enabled: false) changes site
→ Drop-in removed. Object cache off.

get_object_cache_status(site: staging) read-only
→ Drop-in not installed. Backend: Redis configured.

Kimi Code: Off and confirmed. When the clone is done, turn it back on with the same order: status, test, then enable.

How Kimi Code handles this

You are in Kimi Code, about to clone staging into a new test site with your host's tools, and staging uses Redis as its object cache. A clone that starts with the same drop-in and the same connection constants could read and write the same Redis keys as the original, so you want the object cache off on staging first. You ask Kimi Code to turn it off and prove it. It calls get_object_cache_status through the xspeedhub server to see the current state before it proposes anything.

What comes back is the state and the plan: the drop-in is installed and connected, and disabling with toggle_object_cache and enabled set to false removes xSpeed's drop-in and its wp-config.php block, so WordPress goes back to its own per-request cache. A drop-in owned by another plugin would be left alone. The Hub tells the agent to briefly confirm intent for the object-cache toggles, so Kimi Code asks. After your yes it disables, then reads the status again to confirm the drop-in is gone. Turning it back on later follows the usual order: status, test_object_cache, then the toggle with enabled set to true.

What differs in Kimi Code is the naming and the rules. Each Hub tool is mcp__xspeedhub__ followed by its name, and in the default Always Ask mode a call that matches no permission rule asks first. An allow rule for mcp__xspeedhub__get_* covers the status call but not test_object_cache, which starts with test_, so allow that one by name and leave the toggle and the flush unmatched. If the session runs on the server over SSH, Kimi Code can also run redis-cli ping beside the xSpeed test.

Set up Kimi Code once

Already connected? Skip to the prompts. Alternatives and troubleshooting are on the Kimi Code guide.

1Put your sites in xSpeed Hub

Sign in at app.xspeedcache.com with Google or email; the Hub is free and has no site cap. Then connect each WordPress site from its own dashboard: click Connect Hub in the xSpeed Cache top bar, then Connect via xSpeed Hub. Each site needs the free xSpeed Cache plugin.

2Add xSpeed Hub to mcp.json

Put this in ~/.kimi-code/mcp.json to use the Hub in every project, or in .kimi-code/mcp.json inside one repository, where it overrides a user entry of the same name. You can also run /mcp-config and let Kimi Code add it. A server added while a session is open joins only the next session.

~/.kimi-code/mcp.json

{
  "mcpServers": {
    "xspeedhub": {
      "url": "https://app.xspeedcache.com/xspeed/mcp"
    }
  }
}

3Sign in with /mcp-config login

Start a new session with kimi, run this and approve access on the xSpeed Hub page that opens in your browser. There is no token to paste. Then run /mcp to check that xspeedhub is connected.

Text

/mcp-config login xspeedhub
/mcp

1 / 3

Full Kimi Code setup, sign-in options and FAQ

Before you send a prompt that changes something

Reads change nothing on your sites, though contact_support emails xSpeed support. Writes run once your connection allows them, so your client's approval prompt and a read-only connection (the token's Read-only everywhere switch, or a Viewer sign-in) are the gates that matter. In Kimi Code's default Always Ask mode, an MCP call that matches no permission rule asks for approval, and Approve for this session allows the same kind of call until the session ends. Permanent rules go in [[permission.rules]] in config.toml: allow mcp__xspeedhub__get_* and mcp__xspeedhub__list_* to let the reads run, and leave the write tools on the prompt. Ask When Needed mode, from /yolo, approves MCP calls automatically, and Never Ask, from /auto, approves everything. xSpeed Hub has no confirmation step of its own, so in either of those modes a purge or a settings change runs as soon as your connection allows writes. To remove writes altogether, list the read tools in the entry's enabledTools, or use a connection token with Read-only everywhere on.

What do I ask?

Three prompts written for Kimi Code. More for this job are below.

Prompt for Kimi Code

Use xSpeed Hub to turn the object cache off on staging before I clone it, and confirm the drop-in is gone.

Prompt for Kimi Code

With xSpeed Hub, check the object cache on shop and test Redis, then ask me before you enable anything.

Prompt for Kimi Code

This session is over SSH on the web server. Run redis-cli ping, then use xSpeed Hub to test the object cache on blog.

What happens, step by step

A persistent object cache keeps database query results in Redis or Memcached instead of rebuilding them on every request. Through xSpeed Hub an agent can read the state of the cache, probe the cache server, switch the cache on or off, and flush it. The Hub tells the agent to run the probe first, because the status call alone does not prove the server is reachable.

Read the current state

get_object_cache_status reports whether the object-cache.php drop-in is installed, which backend is configured, Redis or Memcached, and its stats. It does not prove the server answers, so a clean status is not a green light.

Test the connection first

test_object_cache opens a live connection to the configured server and does a read and write probe with the settings saved on the site. A failure comes back with the server's own reason, such as a wrong password, a refused connection or a permission error. The Hub tells the agent to run this before toggle_object_cache.

Fix the connection if the test fails

get_settings reads the object-cache module, with secrets masked. update_settings changes it, for example the host or port. On a managed host that defines these values in wp-config.php, the fields are read-only and a write that tries to change one is refused with the name of the constant. For a host that wants the constants pasted, get_object_cache_status with detail set to snippet returns the wp-config.php block.

Turn it on

toggle_object_cache with enabled true makes the site test the connection again, write the connection constants into wp-config.php and install the object-cache.php drop-in. If the test fails, it stops with "Could not enable" and writes nothing. If wp-config.php is not writable, the drop-in goes in and the reply tells you to add the snippet by hand. The Hub tells the agent to briefly confirm intent first.

Check it took

The drop-in loads on the next request, so the agent reads get_object_cache_status again after a page load and runs test_object_cache once more. The status also warns when the cache connects but refuses writes, so a connected cache that stores nothing does not pass as healthy.

Flush when stale data survives

flush_object_cache empties the persistent object cache. Use it when old options, transients or query results outlive a page cache purge. It does not touch the page cache; purge_cache does that.

Reference

Read tools	get_object_cache_status and test_object_cache, both work on a read-only connection
Write tools	toggle_object_cache (enabled true or false) and flush_object_cache
Backends	Redis or Memcached
Status detail	status (default) or snippet, the wp-config.php constants block
What status proves	Drop-in installed, backend configured, stats. Not that the server is reachable
What the test proves	A live connect and a read and write probe on the saved settings, with the server's own error on failure
Enable does	Tests the connection, writes connection constants to wp-config.php, installs the object-cache.php drop-in
Disable does	Removes xSpeed's drop-in and its wp-config.php block; a drop-in owned by another plugin is left alone
Connection timeout	1 second by default, set in the object-cache module
Settings module	object-cache, for get_settings and update_settings
Every site at once	None of these tools accept site: "all"; one site per call
Confirmation	None on the Hub; your client's prompt is the gate

Rules worth keeping

Test before you toggle. The Hub tells the agent to run test_object_cache first, because a cache server that does not answer slows every request while the page cache still looks healthy.
Keep the Redis password out of the chat. get_settings masks it, but the wp-config.php snippet is built from the saved settings and can include a saved password. Set secrets in wp-config.php or in the xSpeed panel.
toggle_object_cache and flush_object_cache are write tools. The Hub tells the agent to briefly confirm intent for the object-cache toggles, but that is guidance, not a lock. Your client's approval prompt and a read-only grant are the gates that matter. A read-only grant means the connection token with Read-only everywhere on, or a Viewer member's sign-in; an OAuth sign-in gets the scopes the client asks for, which are read and write by default.
If another plugin owns the object-cache.php drop-in, turn its object cache off in that plugin. xSpeed leaves a drop-in it did not install in place when you disable.
A flush clears cached data, so the next requests rebuild it from the database and the site is briefly slower. Flush one site, then check it, before you repeat it.

Good to know with Kimi Code

Ask When Needed mode, from /yolo, approves MCP calls automatically, and Never Ask, from /auto, approves everything, so a toggle or a flush would reach the live site with nothing but the Hub's request to confirm in the way. Stay in Always Ask for this job. Approve for this session is fine for the two reads and loose for the writes. For a session that should only check, list get_object_cache_status and test_object_cache in the entry's enabledTools, or use a connection token with Read-only everywhere on.

More prompts for this job

They work in any client connected to xSpeed Hub.

Prompt

Using xSpeed Hub, is a persistent object cache running on shop, and which backend does it use?

Prompt

Use xSpeed Hub to test whether Redis is reachable from shop. Do not change anything.

Prompt

Shop is slow in wp-admin. Use xSpeed Hub to check the object cache before anything else.

Prompt

With xSpeed Hub, test Redis on shop, and if it passes, turn the object cache on.

Prompt

Ask xSpeed Hub for the wp-config.php object cache snippet for docs.

Prompt

Use xSpeed Hub to flush the object cache on blog. Old options survive a page cache purge.

Prompt

With xSpeed Hub, turn the object cache off on shop and confirm the drop-in is gone.

Frequently asked questions

No. Disabling removes xSpeed's drop-in and its wp-config.php block, and WordPress goes back to its own cache for each request. Pages that relied on Redis run their database queries again, so the site may be a little slower until you turn it back on.

Allow mcp__xspeedhub__test_object_cache by name. A rule that matches names starting with mcp__xspeedhub__get_ covers the status call but not the test, because the test tool's name starts with test_. Leave the toggle and the flush without a rule so they ask.

Because a cache server that does not answer costs time on every request while the page cache still looks fine. The Hub tells the agent to run test_object_cache first, and the site also tests the connection when you enable and refuses to continue if it fails. The probe is still worth running on its own, since it returns the server's exact reason, such as a wrong password or a refused connection.

The site keeps working. The drop-in is built never to take the site down: it falls back to WordPress's own cache for the length of the request and marks itself as degraded. A request can still wait up to the connection timeout, which is 1 second by default, so the site gets slower. xSpeed reports that state as degraded, and test_object_cache names the fault.

No. These tools connect xSpeed to a Redis or Memcached server that already exists and switch the object cache on or off. Getting a server is a hosting task. xSpeed uses the phpredis extension when it is there and its own built-in client when it is not.

flush_object_cache empties the persistent object cache and nothing else. purge_cache with type object does the same job through the purge tool, and purge_cache with type all also clears the page cache and, when connected, the Cloudflare edge. Use the flush when stale options or transients survive a page cache purge.

Treat it like a secret. The snippet is built from the settings saved on the site, so it can contain the Redis password if one is saved. Paste it into wp-config.php yourself and avoid sharing it in a chat or a ticket.

Set up Redis object cache, from Kimi Code.

xSpeed Hub is free and has no site cap. Connect your sites once and ask from the agent you already use.

Open xSpeed Hub Kimi Code setup

Set up the Redis object cache with Kimi Code

How Kimi Code handles this

Set up Kimi Code once

What do I ask?

What happens, step by step

Reference

Rules worth keeping

More prompts for this job

Frequently asked questions

Keep going

Set up Redis object cache, from Kimi Code.