Unified finance MCP server: composes DART (structured filings/financials), Toss Securities (execution layer -- prices, orders, portfolio), and Telegram (unstructured news/analyst commentary) into a single MCP server process.
Server has 9 tools with basic descriptions and input schemas, but significant quality gaps. Most tool descriptions are under 100 chars and lack actionable context. Parameter descriptions are minimal or missing. No output schemas documented. No error handling guidance. Tool naming is reasonable (verb_noun pattern mostly followed), but parameter naming inconsistencies (e.g., 'chat' accepting both string and integer without clear guidance on format). No tool annotations (readOnlyHint, destructiveHint). Schemas present but sparse, many parameters lack detailed constraints or format guidance. This is typical community-server quality, landing in the D/C range.
Test DART API connectivity and credentials.
Check DART, Toss, and Telegram credentials/connectivity in one call.
Resolve a Telegram username or numeric ID and return its title, about text, member count, and type.
Fetch recent messages from one Telegram chat, optionally after a date and message offset.
Return dialogs with IDs, titles, usernames, types, and unread counts.
Find Telegram messages carrying documents, photos, videos, audio, or any media, globally or in one chat.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., what fields does get_channel_info return? Is there a member_count field?). This violates pattern:tool and forces agents to guess at response structure.
Parameter 'chat' in get_channel_info, get_messages, search_messages, and search_files accepts both string and integer but lacks guidance on format. Should clarify: 'Telegram username (string, e.g. @username) or numeric chat ID (integer, e.g. 123456789)'. Current ambiguity forces LLM to guess.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
Search message text globally or within one Telegram chat and return matching message summaries.
Test Telegram account connectivity and session validity.
Test Toss Securities API connectivity and credentials.
Test connection tools (finance_test_connection, dart_test_connection, toss_test_connection, telegram_test_connection) have minimal descriptions (under 60 chars). Descriptions should explain: What does 'test' mean? What credentials are checked? What does success/failure look like? Current descriptions lack actionable context.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) present. All 9 tools are READ_ONLY but this is not declared in the tool definition. Agents cannot distinguish safe-to-retry tools from destructive ones without annotations.
Parameter descriptions are sparse or missing. E.g., 'since_date' in get_messages says 'ISO format datetime' but does not clarify: Is it inclusive or exclusive? What happens if the date is in the future? What is the earliest supported date? Vague constraints force LLM to guess.