ChatKeeper vs Combot vs Rose: compare the moderator workflow

If you already have a working moderation bot, replacing it should solve a named problem. Perhaps only one volunteer understands the rules. Perhaps you cannot explain a member's restriction. Or perhaps you need activity reports alongside moderation. Those are better reasons to compare ChatKeeper, Combot and Rose than the size of a feature list.

Our reading of their official documentation suggests three useful starting points: examine Rose for documented command-driven control, ChatKeeper for configurable workflows in its management interface, and Combot when automation and reporting belong in the same evaluation. These are editorial starting points, not rankings or evidence that a feature is exclusive.

What this comparison can establish#

This article is published by Defendy. We reviewed public first-party sources on October 5, 2026; we did not run a cross-product benchmark, inspect paid accounts or measure detection quality. Documentation can establish a described control. It cannot establish your team's learning time, a live deployment's reliability or how a disputed message will be classified.

Sources for these starting points: Rose locks, ChatKeeper instructions and Combot guides and API. The rows do not imply missing capabilities in the other products.

Scene one: a moderator changes a link rule#

Imagine a technical support group that allows documentation links but forbids unrelated invitations. This is a hypothetical evaluation scenario. Give each candidate the same policy, including a permitted documentation URL, a disallowed invitation and an ambiguous recommendation for a third-party tool.

Rose's lock documentation explains content-type controls, allowlists and precedence: locks take priority over blocklists. That is a concrete issue to inspect when two rules seem to disagree. A command list is useful only if the next moderator can find the current configuration and understand which rule wins.

ChatKeeper's automatic moderation guide describes filters and related actions. Follow the same test through its settings: what gets detected, what consequence follows, and what is written to the journal? Do not assume that labels or penalty thresholds map directly from another bot.

Combot's automation guides describe building rules for specific situations. Check the exact condition and action you need in the current interface. A rule builder can reduce repetitive work, but a complicated rule still needs a plain-language explanation and an owner.

For all three, ask the evaluator to change one item, reopen the configuration and repeat the allowed case. Record the result and the changed setting. “I clicked Save” is weaker evidence than a persisted setting and the expected behavior.

Scene two: a legitimate newcomer gets stuck#

A newcomer check should have an understandable failure path. Test a successful join, an abandoned challenge and a person returning after an interrupted attempt. Use an ordinary consenting test member, not only the owner account.

Rose's CAPTCHA guide documents several flows and notes that someone who previously completed verification may not be challenged again. That matters when preparing a repeatable test. ChatKeeper's newcomer guide connects greeting, verification and join-request settings. Test the whole entrance path rather than an isolated button.

For each candidate, identify who can help a blocked person, where that person can contact them and whether recovery requires rejoining. Do not credit any product with accessible onboarding merely because “CAPTCHA” appears in its feature list. Language, mobile navigation and timeout behavior need your own check.

Scene three: an appeal arrives the next morning#

The moderator on duty did not see the original incident. Can they determine which rule acted, what restriction remains and whether another bot was involved?

Rose documents log categories. ChatKeeper's journal guide explains event logging and distinguishes a command that returns recent records privately from one that posts them in the group. Combot's API reference includes a moderation journal endpoint. These are different retrieval paths, not proof of identical coverage or retention.

Keep the recovery test narrow: review the event, remove the mistaken restriction, correct the rule or exception and retest. Do not promise to restore a deleted Telegram message. Preserving context securely before changing settings is often more useful than adding another detector.

Handover, permissions and leaving later#

Ask a second moderator to operate the test group without the original configurator present. Give them a short policy document, official documentation and their intended role. The exercise should reveal missing access and hidden assumptions before a real incident does.

Then check the exit route. Rose documents JSON settings exports and owner-only imports. That is not a guarantee that another vendor can import the file. For all products, distinguish configuration export, event export and participant data; ask about each required category rather than accepting “export supported”.

Review bot permissions and dashboard access separately. Do not grant administrator-promotion rights as a convenience. Avoid running two enforcement bots on the same rules during a trial: a deletion by one can prevent a meaningful observation of the other.

Commercial terms also need a separate check. Confirm the required plan, group scope, renewal, data retention and downgrade behavior using the current offer. The paid-versus-free guide provides a budget and exit checklist without assuming that a paid tier is necessary.

Where Defendy belongs in this decision#

Defendy is another candidate if per-group configuration inside Telegram is important to you. A separate implementation review confirmed named-list management in its Mini App; see the stop-word guide for navigation. This is code evidence, not a live-service availability or performance claim; public rollout and account-specific pricing were not established by this review.

Apply the same scenarios to it. The useful outcome of a comparison is a team that can explain its choice, operate the rules and recover from mistakes. Keeping your current bot after fixing a handover problem is also a valid outcome.