# Keepcon MCP MCP > Keepcon automates content moderation and semantic analysis for user-generated content. It lets your AI agent instantly check text for policy violations, tag content by category, and manage large queues of submissions. You can process texts in real time or import massive batches for background review. ## Overview - **Category:** customer-support - **Price:** Free - **Tags:** content-moderation, semantic-analysis, user-generated-content, automated-tagging, customer-experience ## Description You need to moderate the constant stream of comments, forum posts, and messages without hiring a full-time team just for monitoring. This MCP connects your AI client directly to Keepcon's moderation engine. Your agent checks text against policy guidelines, giving immediate 'approve/reject' decisions and semantic tags. If you have hundreds or thousands of submissions, the tool handles batch processing asynchronously, letting you retrieve results later. You can also manage the entire lifecycle by listing user profiles and submitting feedback to improve the system’s accuracy over time. It all connects through Vinkius, giving your agent a single point of access for governance tasks. ## Tools ### acknowledge_results Marks moderation results as received, clearing them from the pending queue. ### export_results Retrieves all moderated content that was submitted in a large batch. ### submit_feedback Sends moderation feedback, such as identifying false positives, to improve the system's accuracy. ### get_profile_by_social_id Gets a user profile when you only have their social media handle (like Twitter or Facebook). ### get_profile Fetches a user's profile details using their internal Keepcon ID. ### import_batch Submits large amounts of content for moderation, which returns an ID for tracking. ### moderate_content Checks single pieces of text immediately for approval status and semantic tags. ### list_profiles Fetches a list of all user profiles available in your account. ### search_profiles Searches the user database using specific filters to find targeted profiles. ## Prompt Examples **Prompt:** ``` Moderate this text in the 'forum' context: 'This user is being very aggressive!' ``` **Response:** ``` I've processed the text. Keepcon's decision is 'REJECT' with the tag 'Aggressive Behavior'. Would you like to see more details? ``` **Prompt:** ``` Export pending moderation results for the 'chat' context. ``` **Response:** ``` I've retrieved 15 pending results for context 'chat'. 12 were approved and 3 were rejected. The result set ID is 'SET_98765'. ``` **Prompt:** ``` List all user profiles in my Keepcon account. ``` **Response:** ``` I found 50 user profiles. Notable users include 'user_abc' (Twitter) and 'user_xyz' (Facebook). ``` ## Capabilities ### Moderating live content Send text immediately to get an approval decision and category tag. ### Processing bulk submissions Submit huge volumes of content for background moderation and retrieve the results later. ### Managing moderation queues Export pending decisions or mark processed batches as acknowledged to keep your queue clean. ### Identifying users List, search for, and retrieve specific user profiles using their social or internal IDs. ### Improving accuracy Submit feedback on moderation decisions to help the system learn and become more precise. ## Use Cases ### The forum thread cleanup A community manager gets a new batch of posts. Instead of reading them, they ask their agent to `import_batch` the content. The agent runs the moderation and then uses `export_results` to pull down all 50 rejected items for manual review. ### Filtering competitor spam A security team needs to check a user's history. They use `search_profiles` with specific filters, then use `get_profile` on the resulting IDs to validate if that account is associated with known bad actors. ### Testing new content guidelines A developer wants to see how a piece of text performs. They call `moderate_content` first for an instant decision, then they use `submit_feedback` if the result is wrong, helping refine their own policy before full deployment. ### Onboarding new staff A team lead needs to verify all active user accounts. They run `list_profiles` to see everyone, then use `get_profile_by_social_id` for key users to check their status across different social networks. ## Benefits - Instant decisions: Use `moderate_content` to check any text in real time. Your agent gets an immediate approve/reject decision and a tag, so you don't wait for a batch process. - Handle scale with confidence: When dealing with thousands of posts, run `import_batch`. This processes the content offline, letting you pull results later using `export_results` without timing out. - Maintain data hygiene: After processing a large set of items, use `acknowledge_results` to mark them as handled. This keeps your moderation queue accurate and prevents repeat work. - Profile visibility: Need to know who posted something? Use `get_profile_by_social_id` or `search_profiles` to pull up user details using just their social media handle. - System improvement: The tool lets you submit feedback via `submit_feedback`. This isn't just moderation; it actively trains the semantic engine to be better over time. ## How It Works The bottom line is you get an automated feedback loop: submit content, check decisions, manage results, and improve the model using the same agent workflow. 1. Subscribe to this MCP, then enter your Keepcon API Key and Account Number. 2. Your agent calls a tool like `import_batch` to start moderating content or runs `moderate_content` for real-time checks. 3. The system returns the moderation decision (approve/reject) and tags. You can then use `export_results` to pull down the final, bulk results. ## Frequently Asked Questions **How do I use Keepcon MCP for real-time moderation?** You run `moderate_content` directly through your agent. It takes text as input and immediately returns a decision (approve/reject) along with any semantic tags it finds. **Is there a way to moderate huge amounts of posts using Keepcon MCP?** Yes, use `import_batch`. This sends your content for background moderation and gives you an ID. You then wait and retrieve the full results later with `export_results`. **Can I link a user profile to their moderated posts using Keepcon MCP?** You can use `get_profile` or `search_profiles`. This lets your agent pull up specific user details, which you can then cross-reference with the content that was flagged. **What if I think a moderation decision is wrong? How do I fix it using Keepcon MCP?** You use `submit_feedback`. This tool lets your agent send detailed feedback, like flagging false positives. The system uses this input to retrain its semantic engine and improve accuracy over time. **What credentials do I need to connect Keepcon MCP?** You must provide a valid Keepcon API Key and an Account Number. These keys establish your connection rights and prove you have access to the service. **After running batch moderation, how do I clear out old results using `acknowledge_results`?** `acknowledge_results` lets you confirm receipt of processed data. Running this tool clears pending records from the queue, keeping your system clean and accurate for future runs. **How can I find a user profile if they don't have an associated forum account? Does Keepcon MCP support social IDs?** Yes, use `get_profile_by_social_id`. This tool accepts external identifiers like Twitter or Facebook IDs to pull up the correct user profile record. **If I need to find a specific type of user among thousands, is there a better way than listing all profiles?** You should use `search_profiles`. This tool lets you apply specific filters, which narrows down the results set and helps you quickly pinpoint the exact users you're looking for. **What is the difference between synchronic and asynchronic moderation?** Synchronic (`moderate_content`) provides an immediate decision, while asynchronic (`import_batch`) is for large volumes where results are retrieved later via the export tool. **How do I ensure results are not exported twice?** After retrieving results with `export_results`, use the `acknowledge_results` tool with the corresponding `set_id` to confirm processing. **Can I provide feedback on incorrect moderation decisions?** Yes, use the `submit_feedback` tool to report false positives or negatives, helping the Keepcon engine learn and improve over time.