tutorials

Cover Five Platforms With One Manus Cue Connector for Social Listening

Taras Shynkarenko
Taras Shynkarenko
•Updated: •7 min read
One Manus Cue connector for social listening covering Reddit, X, Bluesky, Hacker News and FacebookOne Manus Cue connector for social listening covering Reddit, X, Bluesky, Hacker News and Facebook

TL;DR

7 min read

Cue, the Manus agent app launched on 28 September 2026, ships no connector for Reddit, X, Bluesky, Hacker News or Facebook. One custom MCP connector to RedReplier returns all five in a single list_mentions call, scored and tagged by source. Count per source before reading, sort by RELEVANCE, keep includeLowRelevance off, and give the connector to one agent, because Manus has published nothing on how agents in a group chat share credentials.

Reddit, X, Bluesky, Hacker News and Facebook each need their own login, rate limit and page layout, which is the case for running a Manus Cue connector for social listening through one monitoring API instead of five. Cue is the personal agent app Manus launched on 28 September 2026, and its launch material names no connector for any of the five. One custom MCP connector to RedReplier gives a Cue agent all of them, merged and scored before the agent reads a line.

The short version: add RedReplier's MCP server as a custom connector, have the agent count mentions per source before it reads any, sort by RELEVANCE, and keep the connector on one agent. Every platform arrives through the same list_mentions call.

Can one Cue connector cover Reddit, X, Bluesky, Hacker News and Facebook?

One RedReplier connector covers all five platforms, because GET /mentions returns them in a single response with the platform as a field on each mention. The MentionSource enum in RedReplier's spec has seven values: REDDIT_POST, REDDIT_COMMENT, TWITTER, BLUESKY, HACKERNEWS, FACEBOOK and FACEBOOK_GROUP. Reddit and Facebook each split into two.

The connector goes in through the Manus settings path, Settings, then Integrations, then Custom MCP Servers, with the URL https://mcp.redreplier.com/mcp and a redreplier_ Bearer token. As of 29 September 2026 Manus has not documented whether a connector added there appears in Cue, so ask the Cue agent to list your monitored websites as a test. The Manus Cue connector for Reddit monitoring pillar walks through both routes and the token role.

A person at a desk switching between several screens, showing the effort of reading many separate feeds one at a time.

Why use a monitoring API instead of letting the agent browse?

A monitoring API hands the agent a list that is already matched and scored, while browsing hands it five websites to read one page at a time. Manus gives each Cue agent its own computer to "see a task through on its own machine," and Manus says its general API calling reaches any service with a well-documented REST API. Neither turns five feeds into one ranked list.

Five platforms, two ways in
The agent browses five sites
  • Five logins or five public views
  • Five layouts that change without notice
  • Five sets of terms and rate limits
  • No score, so the agent judges every post
One connector to RedReplier
  • One MCP URL, one Bearer token
  • relevanceScore on each mention
  • NEW, APPROVED and REJECTED carry across runs
  • Seven source values in one call
The right column is the argument. The matching and scoring happen before the agent reads anything.

X is the expensive platform to go direct on, because holding your own X API access means paying X for every post read, which our X API pricing page breaks down. RedReplier's TWITTER mentions come through the same token as the rest.

Status is the part people underrate. Each RedReplier mention carries NEW, APPROVED or REJECTED, so tomorrow's run skips what you dealt with today. A browsing agent keeps no such record unless you build one.

Hands sorting papers into separate piles, illustrating how counting and ranking mentions by source keeps a digest manageable.

How do I stop Cue reading hundreds of mentions?

Cue reads fewer mentions when it counts before it lists and sorts by relevance. count_mentions returns one total for whatever filters you pass, so call it once per source first and you see the skew before it floods the digest. Then list_mentions with sort=RELEVANCE puts a Hacker News thread scored 90 above an X post scored 30, whenever each one landed.

You have a RedReplier connector. You listen for me across platforms.
 
Allowed tools: list_websites, list_mentions, count_mentions,
explain_mention. Never create, edit, delete, enable, disable,
approve or reject anything. Never change alert settings.
 
Each morning:
1. count_mentions for NEW mentions from the last 24 hours, once per
   source: REDDIT_POST, REDDIT_COMMENT, TWITTER, BLUESKY, HACKERNEWS,
   FACEBOOK, FACEBOOK_GROUP. Show me the seven counts first.
2. list_mentions for the same window, HIGH and VERY_HIGH only,
   sort=RELEVANCE, at most 5 per source.
3. Put mentions with no score in their own short group.
4. Explain the top three overall.
 
Never contact anyone who appears in a mention. Report in this chat.
If every count is zero, send nothing.

Two defaults work in your favour. includeLowRelevance is false unless set, which leaves out mentions the scorer judged unlikely to deserve a reply. And limit defaults to 50 with a ceiling of 500, so the per-source cap in step 2 is the agent's job, not the API's.

Does listening belong in one Cue agent or a group chat?

Social listening runs best on one Cue agent that holds the connector, with a group chat only for work that needs no RedReplier access. Manus describes group chats where agents "hand work to each other," and a listening setup maps onto that: one agent reads RedReplier, another turns a week of digests into a summary for your team.

As of 29 September 2026 Manus has published nothing on isolation between Cue agents or on whether agents in a group chat share a connector. Manus's docs say that "integration connections are personal to your account," which describes Manus, not how Cue divides access between your agents. Test it: ask the summary agent to list your RedReplier websites. If it answers, it holds the connector too and needs the same brief.

Checking whether a second agent holds the connector
Ask the summary agent to list your RedReplier websites
It answers with your websites
It holds the connector too
Give it the same brief
Manus has published nothing on this, so the test is the only answer.

Which calls change what you listen to?

RedReplier's keyword and website calls change what the connector hears, and none belongs to a listening agent. add_keywords, edit_keyword, delete_keyword, disable_keyword and enable_keyword change what gets matched, and update_website rewrites the description that "feeds mention scoring," per the spec. A listening agent that tunes its own keywords drifts from the conversations you asked it to follow.

delete_keyword is the worst of them, because RedReplier's agent docs say it removes "a keyword and the mentions it found." disable_keyword pauses matching and keeps the history. Ask the agent to suggest keyword changes in the digest, and make them yourself in RedReplier.

RedReplier
RedReplier

Get Started

Reddit, X, Bluesky & HN

Real-time intent alerts

Unlimited AI replies

Ranked by buyer intent

What breaks across five platforms?

Four things break once five platforms share one agent, and each has a fix you write into the brief.

Four breaks and the fix for each
1
Null relevanceScore. The spec marks the score nullable. Return unscored mentions as their own group so a numeric sort does not bury them.
2
Null url. Some Facebook and Bluesky posts have no public link. Tell the agent to say so instead of searching for the author.
3
Split sources. Reddit and Facebook each use two source values. A per-source cap that lists only REDDIT_POST lets every comment through.
4
Source skew. Reddit and X out-produce the others. Count per source first, then cap per source.
Every fix lives in the brief, because Manus documents no approval step inside Cue as of 29 September 2026.

The null url case matters more on Cue than on other agents. An agent with its own email and phone that cannot find a link has the means to go looking for the author. The contact ban in the brief closes that route, and the lead finding guide covers why it matters most when the mention is a buyer.

The token breaks too, in one specific way. RedReplier checks redreplier.write on every call, reads included, so a Viewer key returns 403 permission_denied on the first count_mentions. Create the key under the Editor role at redreplier.com/api-tokens with an expiry date.

How frequently does RedReplier check the five platforms?

RedReplier monitors in real time, per its pricing page, and your plan sets the email alert cadence. GET /alert-settings returns cadenceMinutes next to minIntervalMinutes, the fastest cadence the plan allows, and PUT /alert-settings clamps a faster request to that floor. The Cue agent checks whenever it runs, which is separate from the email cadence.

Whether a Cue agent runs on a timer is undocumented as of 29 September 2026. Manus 2.0 Automations run on a schedule or on triggers including Gmail and RSS, which the Reddit monitoring pillar covers as the fallback route.

Frequently Asked Questions

Does one connector cover all five platforms?

One RedReplier connector returns Reddit, X, Bluesky, Hacker News and Facebook through the same list_mentions call. The platform is a source field on each mention, with seven values in total. Narrowing to one platform is a filter change, not a second connector.

Can Cue post replies on these platforms through RedReplier?

RedReplier has no endpoint that publishes a reply on any platform. Each mention carries an aiReplySuggestion, a "draft reply for a human to review before posting." Keep the agent's own email and phone out of it with a contact ban in the brief.

Why sort by relevance instead of recency?

sort=RELEVANCE ranks by score across platforms, while sort=RECENT ranks by publish time. Reddit and X produce more matches per hour than Hacker News or Bluesky, so a time-sorted digest fills with the two busiest sources. Relevance lets a slow platform's best thread reach the top.

Does Cue work with RedReplier's social listening on mobile?

Manus lists Cue for web, desktop and mobile, with the iOS app waiting on App Store review as of 29 September 2026. Manus has not documented whether a connector behaves differently between Cue clients, so run the websites test on each one you use. Access needs an invite code, such as MEETCUE, for a limited number of early users.

How does this compare with the Muse and Grok versions?

The Meta Muse connector for social listening builds a REST client from RedReplier's spec, and the Grok bot template for social listening ships the brief pre-written. Cue takes an MCP server instead. It also gives the agent an email and phone number, which is why this brief carries a contact ban.

Where do I see RedReplier's listening without an agent?

RedReplier's social listening tool page covers listening on Reddit: tracking topics rather than bare name-drops, reading sentiment in context, ranking threads by reply opportunity, and reviewing the full thread with a suggested angle before you reply. The Cue connector reads the same mentions the dashboard shows. Changes you make in the dashboard, such as approving a mention, show up in the agent's next list_mentions call.

What role does the RedReplier token need for a Cue connector?

RedReplier checks redreplier.write on every call, reads included, so the token needs the Editor role. A Viewer key returns 403 permission_denied on the first count_mentions. Create the key at redreplier.com/api-tokens and set an expiry date.

Which RedReplier tools should a listening agent be allowed to use?

Four: list_websites, list_mentions, count_mentions and explain_mention. The keyword calls change what gets matched, and delete_keyword removes the keyword and the mentions it found. Name the allowed tools in the brief and forbid the rest.

What should Cue do with mentions that have no relevance score?

The spec marks relevanceScore as nullable, so some mentions arrive unscored. Have the agent return them as their own short group, so a numeric sort does not bury them.

How do I check that Cue can see the RedReplier connector?

Ask the Cue agent to list your monitored websites. As of 29 September 2026 Manus has not documented whether a connector added under Custom MCP Servers appears in Cue, so the answer tells you. If it lists them, the connector works.

RedReplier
RedReplier

Get Started

Reddit, X, Bluesky & HN

Real-time intent alerts

Unlimited AI replies

Ranked by buyer intent

Create a RedReplier API token with the Editor role, connect it to one Cue agent, and ask for tomorrow's seven source counts before any mention text.

Sources: Manus's Introducing Manus 2.0 (28 September 2026), How can I use Manus Connectors, the custom MCP docs, integrations docs and Automations docs, the RedReplier OpenAPI spec and agent docs. Checked 29 September 2026.

See us more often in Google

One click marks RedReplier as a preferred source, so our articles sit higher in your Top Stories, AI Mode, and AI Overviews.

Before you go...

RedReplier

RedReplier

Catch every buyer asking for what you sell

RedReplier watches Reddit, X, Bluesky and Hacker News in real time, ranks every thread by buyer intent, and drafts your reply, so you get there first.

Reddit, X, Bluesky & HN

Real-time intent alerts

Unlimited AI replies

Ranked by buyer intent

Related Articles