Mattermost should separate requireMention for channel vs thread messages

A requested OpenClaw Mattermost behavior change would let operators require mentions in channels without forcing the same rule inside threads.

Why this matters

In many Mattermost setups, channel traffic and thread traffic serve different purposes. Channel messages are noisy and may need a mention gate, while thread replies often represent already-engaged conversations where mention requirements become friction.

That means a single global requireMention policy can create the wrong kind of consistency: it is technically simple, but operationally confusing.

What operators actually want

The requested behavior is simple:

require mentions for top-level channel messages
but allow replies inside an active thread without forcing another explicit mention

That split reduces channel noise without making thread follow-ups feel broken.

Common failure pattern

When a single requireMention rule applies everywhere, operators can see this pattern:

the bot correctly ignores unmentioned top-level posts
but it also ignores thread replies that users reasonably expect it to answer
users read that as thread unreliability rather than as policy
the support team starts checking delivery logs even though the message reached the bot layer

This is why the issue has a high troubleshooting intent. It looks like message loss, but it may actually be configuration granularity.

How to confirm you hit the same problem

Before treating this as the same issue, confirm all of the following:

the Mattermost integration is otherwise healthy
the bot responds when explicitly mentioned
the missing replies are concentrated in thread follow-ups
the behavior changes when users add a fresh mention inside the thread

If those checks line up, the problem is likely policy granularity rather than a transport or webhook outage.

What this is not

Rule out these simpler explanations first:

the Mattermost bot was offline or disconnected
the channel binding itself was misconfigured
the user reply landed outside the expected thread context
the message was filtered by a different permission or mention rule
the bot never received the event at all

If one of those is true, you are looking at delivery or routing, not mention-policy scope.

Why teams misdiagnose it

This pattern is easy to misread because the first message and the follow-up message happen in the same visible conversation. Humans assume the bot should inherit that context automatically.

But if policy is evaluated the same way for both top-level and thread traffic, the thread reply can still be rejected even though the conversation already feels active to the user.

Temporary operator guidance

Until the config model separates channel and thread mention policy, teams can:

document that thread replies still need explicit mentions
reserve bot-heavy workflows for channels where mention behavior is predictable
avoid assuming thread context alone will trigger a response
train moderators to distinguish “ignored by policy” from “lost in transport”

The most practical workaround is simple: if the bot answered once in a thread, do not assume later replies will work without another mention unless you have verified that behavior in your own deployment.

Recovery and response playbook

If users report that the bot “stops answering in threads,” use this order:

verify the bot still answers a new explicitly mentioned message
compare top-level channel behavior against thread behavior
test whether a fresh mention revives the reply path
only then escalate to logs or delivery debugging

That sequence prevents wasted time on the wrong layer.

Source

Candidate issue tracked in the internal editorial queue