Campaign Swarm
Server Details
Campaign Swarm is an AI campaign workspace for agencies. Plan campaigns, create content, schedule posts, review performance, and manage approvals from your MCP client.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 35 tools
Descriptions are unusually explicit and do a lot of work, but the core vocabulary collides: create_post_draft actually creates a content *plan*, while create_planner_campaign also creates a plan and create_content_item creates the asset that sounds like a 'post draft'. Similarly list_plans vs list_planner_campaigns and get_planner_campaign vs get_plan_and_usage invite misselection without reading the fine print. Lifecycle tools (get_planner_stage/quality/retry) are cleanly separated.
Overwhelmingly consistent verb_noun snake_case (get_*, list_*, create_*, update_*, start_*, retry_*). Minor deviations: reply_to_comment and schedule_post_from_draft insert prepositions, and apply_performance_learnings reads as a domain phrase rather than an action on a named resource. Predictable enough for an agent to guess names.
35 tools is heavy, though the surface spans genuinely distinct sub-domains (projects/plans, assets, publishing/queue, schedules, comments, ads, analytics, 12-stage planner), so roughly 5-8 tools per area. Still, several clusters could be folded (planner get/stage/quality) and the breadth raises selection cost. Borderline heavy rather than obviously over-scoped.
Read/create/update coverage is strong across planning, assets, scheduling, ads and analytics, and the deliberate human-approval boundary is coherent. But there is no delete or cancel operation anywhere: no way to remove a project, plan, or content item, and no tool to cancel a queued row even though list_publish_queue tells agents a human can cancel it. Ads are read-only plus propose-pause/budget, so campaign creation is absent.
Available Tools
35 toolsapply_performance_learningsApply what works to a planAInspect
Write the workspace's current measured findings (engagement rate, CTR, format, slot, hook) into a plan's brief, so every asset that plan produces next is briefed from evidence. Refuses below the minimum sample instead of inventing a pattern. This changes the brief; it does not regenerate existing assets.
Inputs: plan_id (required), days.
create_content_itemWrite an asset onto a planBInspect
Create a copy/image/email asset with the text you supply, attached to an existing plan. This is the override path: the planner does not write the copy, but the asset still inherits the plan's brief. Nothing is published.
Inputs: caption (required), plan_id (required), type, status, +2 more.
create_planner_campaignCreate a strategy campaignAInspect
Create a 12-stage Campaign Studio plan from a brief. It does not start or spend tokens until start_planner_campaign is called. Human approval and publishing remain outside MCP.
Inputs: name (required), intake_data (required), mode, description.
create_post_draftCreate a post draftBInspect
Create a content plan so the swarm can produce assets for it. Nothing is published by this tool.
Inputs: name (required), objective (required), strategy_campaign_id (required), platforms, +2 more.
create_projectCreate a brand projectBInspect
Create a project (a product or client) and optionally link its social accounts and brand brief. Content plans written for this project are grounded in these facts; it does not create or publish content.
Inputs: name (required), brand, website, channel_ids.
execute_content_planRun the content swarmAInspect
Execute a linked content plan through the real content swarm: planner brief, copy, image, video storyboard, email and compliance review. This spends AI credits and creates draft/approved assets; it does not publish anything.
Inputs: plan_id (required).
get_ad_campaign_detailsAd campaign detailsCRead-onlyIdempotentInspect
One ad campaign with pacing, spend, ROAS and current state.
Inputs: campaign_id (required).
get_ads_overviewAds overviewBRead-onlyIdempotentInspect
Meta ads spend, ROAS, campaign count and pending approvals for the workspace.
Inputs: no named parameters.
get_content_assetsContent assetsCRead-onlyIdempotentInspect
Assets the swarm produced (image, video, copy, email) with their plan, status and prompt.
Inputs: limit, plan_id.
get_data_catalogueData catalogueBRead-onlyIdempotentInspect
The datasets this workspace can ground a plan in — sources, row counts, and the facts the agents are briefed with.
Inputs: no named parameters.
get_page_summaryPage summaryBRead-onlyIdempotentInspect
Followers, reach, impressions and profile views for one account over a window.
Inputs: account_id (required), days.
get_performance_digestWhat has workedARead-onlyIdempotentInspect
Measured post performance with sample sizes: findings, top posts, plan deltas and the prompt block content generation uses. Read this before proposing new content so the advice is grounded in this workspace's numbers.
Inputs: days, plan_id.
get_plan_and_usagePlan, limits and usageARead-onlyIdempotentInspect
The workspace plan, every limit it sets and what has been used this window (requests/min, posts and AI credits per month, brands, profiles, seats). Read this before telling a user they cannot do something — a 429 names the metric it refused; this names the reset time and the plan that would allow it.
Inputs: no named parameters.
get_planner_campaignGet a strategy campaignARead-onlyIdempotentInspect
One 12-stage plan with stage statuses and a quality summary. Use get_planner_stage for a stage artifact and get_planner_quality for the gate verdict.
Inputs: campaign_id (required).
get_planner_qualityGet the quality gateARead-onlyIdempotentInspect
The blocking quality gate for a strategy campaign: passed, score, launch recommendation and violations. A campaign is not launch-ready unless passed is true; this tool never approves it.
Inputs: campaign_id (required).
get_planner_stageGet a stage artifactBRead-onlyIdempotentInspect
The full result of one stage: segmentation, personas, brief, journey, outlook, creative, quality_gate, simulation or monitoring. This is how an agent reads the planner's actual output.
Inputs: stage (required), campaign_id (required).
get_post_performancePost engagement curveARead-onlyIdempotentInspect
Every metric capture for one published post (1h/24h/72h/7d/30d), attributed to its plan and content item. Platforms without a measurable API return status 'unsupported' rather than a zero.
Inputs: queue_id (required).
get_swarm_activitySwarm activityCRead-onlyIdempotentInspect
What the agents have been doing — the inter-agent transcript.
Inputs: limit.
get_top_postsTop published postsCRead-onlyIdempotentInspect
Published posts with their reach and engagement metrics.
Inputs: limit, account_id.
list_commentsList commentsBRead-onlyIdempotentInspect
Comments and mentions awaiting a reply, with their status (new, replied, hidden, ignored).
Inputs: limit, status, account_id.
list_planner_campaignsList strategy campaignsCRead-onlyIdempotentInspect
The 12-stage Campaign Studio strategy runs in this workspace, with status and current stage. Read this before creating content so the plan is linked to a real strategy.
Inputs: limit.
list_plansList content plansCRead-onlyIdempotentInspect
Content plans (the basis of every asset) with their status and grounding.
Inputs: limit.
list_projectsList projectsARead-onlyIdempotentInspect
Projects (a product or client) that scope chats, brand profiles and content. Includes the brand brief agents are briefed from.
Inputs: no named parameters.
list_publish_queueList the publish queueBRead-onlyIdempotentInspect
Scheduled and published posts with approval state, account, time and content item. Read this before scheduling so a new item does not collide with an existing queue row.
Inputs: limit, status, plan_id, platform, +1 more.
list_schedulesList autopilot schedulesBRead-onlyIdempotentInspect
Recurring schedules that post a plan's content: cadence, timezone, approval mode, next run and how many runs so far.
Inputs: no named parameters.
list_social_accountsList connected accountsBRead-onlyIdempotentInspect
Connected Facebook pages, Instagram business accounts and Threads profiles, with tracking state and last sync.
Inputs: no named parameters.
pause_ad_campaignPause an ad campaignAInspect
Propose pausing an ad campaign. The change is queued for approval, not applied silently.
Inputs: reason (required), campaign_id (required).
reply_to_commentReply to a commentBInspect
Post a reply to a comment on one of your posts. Creates the reply on the platform and records it in the inbox.
Inputs: text (required), comment_id (required).
retry_planner_stageRetry a failed stageAInspect
Retry a stage that failed, including a quality_gate that blocked the campaign. Only failed stages can be retried; this cannot set a status or force a pass.
Inputs: stage (required), campaign_id (required).
schedule_post_from_draftSchedule a draftAInspect
Put an approved asset in the publish queue for a given account and time. Creates a queue item a human can still cancel.
Inputs: content_item_id (required), social_account_id (required), scheduled_for.
start_planner_campaignStart a strategy campaignAInspect
Start the 12 stages for an existing draft. This spends model tokens. It does not approve a stage, bypass the quality gate, or publish anything.
Inputs: campaign_id (required).
update_ad_budgetPropose an ad budget changeBInspect
Propose a budget change (max ±30% per call, reason required). It becomes an approval item — it never changes the budget silently.
Inputs: reason (required), campaign_id (required), daily_budget (required).
update_content_assetEdit an assetAInspect
Change an existing asset's caption, hashtags or review status (draft/approved/rejected). Use this to correct generated copy instead of regenerating it.
Inputs: item_id (required), status, caption, hashtags.
update_plan_briefUpdate a plan briefAInspect
Rewrite the brief a plan's agents work from: objective, audience, platforms, and the dense context pack (summary/position/claims). This changes what the planner would produce next; it does not regenerate assets on its own.
Inputs: plan_id (required), name, claims, objective, +4 more.
update_scheduleEdit, pause or resume a scheduleAInspect
Change an autopilot schedule: cadence, times, days, timezone, approval mode, run cap, stop rule, or status (active/paused/finished). Editing re-arms the next slot; pausing stops future slots without touching posts already queued.
Inputs: schedule_id (required), name, times, status, +7 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceRemote MCP server that lets AI assistants read tracked competitive intelligence — competitor news, reviews, website changes, social posts and ads — and act on it by managing projects, competitors, battle cards, reports and canvases. Every answer carries a source and a date, and write actions only publish when explicitly shared.MIT
- AlicenseCqualityCmaintenanceCompetitor Monitor AI - MCP server providing AI-powered tools and automation by MEOK AI Labs117 npm46 PyPIMIT
- AlicenseAqualityDmaintenanceEnables brand visibility monitoring across major AI platforms like ChatGPT, Claude, Gemini, and Perplexity. It allows users to track visibility scores, analyze competitor data, and receive actionable insights to improve AI-generated brand recommendations.169 npm1MIT

industrylens-mcpofficial
AlicenseNot gradedqualityCmaintenanceBrowse IndustryLens's published competitive-intelligence reports and head-to-head competitor comparisons from any AI agent — real, source-backed data.MIT
Glama MCP Gateway
Add one secure layer between your agents and this server.