Mail.ru MCP Assistant Starter
Securely search, read, and organize Mail.ru messages using a self-hosted MCP server.
Free
Works with the AI Agents you already use
Mail.ru MCP Assistant Starter
Example session with this skill installed
Find the recent email from 'GitHub' about security alerts and summarize the content.
- Read your context and instructions
- Compiled the mail ru mcp
I have searched your Mail.ru inbox for recent messages from GitHub regarding security alerts. I found one relevant message in your Inbox.
Search Results & Summary
| Date | Sender | Subject | Folder | UID |
|---|---|---|---|---|
| 2023-10-24 | GitHub noreply@github.com | [GitHub] Security alert for your repository | INBOX | 4521 |
Message Summary (UID: 4521):
The email notifies you of a Dependabot security alert regarding a high-severity vulnerability found in one of your dependencies. It recommends updating to the latest patched version of the package to mitigate potential risks. The message includes a link to the specific repository alert page for further details.
Safety Note: I have extracted the text but did not follow any links contained in the message body.
Next steps
- Provide the UID if you would like me to move this message to a specific folder.
- Ask me to search for other alerts from specific dates if needed.
- Type "Move UID 4521 to Trash" if you wish to remove this notification.
Connects securely to your tools. The creator never sees your data.
About this skill
Managing a Mail.ru inbox through an LLM requires strict safety and precise tool execution. This skill provides a structured workflow for interacting with a self-hosted Mail.ru MCP server to handle message retrieval and organization. It eliminates the friction of manual IMAP searching by translating natural language queries into specific MCP tool calls.
What it does
- Folder discovery calls
mail_list_foldersto map the mailbox structure and locate specific archives or labels. - Targeted searching uses
mail_searchwith defined filters to isolate messages by sender, date, or subject without scanning the entire inbox. - Secure reading invokes
mail_readto fetch content while treating all body text and links as untrusted data to prevent prompt injection. - Verified organization utilizes
mail_moveandmail_deleteto reorder or trash messages based on unique UIDs. - State verification checks source and destination folders after every mutation to prevent duplicate actions or sync errors.
How it works
- The agent identifies the mailbox goal and checks for the presence of the five core MCP tools.
- It executes discovery or search commands to find the specific message metadata required for the task.
- For destructive actions like moving or trashing, it surface the UID and folder for explicit human authorization.
- It performs the tool call and runs a follow-up check to confirm the server state matches the requested change.
Frameworks & tools
This skill requires a deployed instance of the mailru-mcp-starter-free server. It utilizes standard IMAP-based tools including mail_list_folders, mail_search, mail_read, mail_move, and mail_delete via the Model Context Protocol (MCP).
Why this beats prompting it yourself
Generic prompts often fail to handle UID-based targeting, leading to the wrong emails being moved or deleted. This skill enforces a verification loop and safety check that prevents the agent from following malicious instructions hidden inside email bodies.
Use cases
- Find and summarize specific travel itineraries or receipts from a cluttered inbox.
- Clean up specific newsletters by moving identified senders to the Trash.
- Check for unread messages within specific subfolders without changing their read status.
Known limitations
Requires a user-deployed MCP server. Does not support bulk deletions, permanent purging, sending emails, or emptying the Trash folder.
Requires a self-hosted Mail.ru MCP server. No support for sending mail, bulk actions, or permanent deletion. Human approval required for all mutations.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
This skill has been used 1 time.
Trust & safety
Security scan
Security scan passed 7 days ago
- Free with an Agensi account
Requires