Firebird 데이터베이스를 위한 Anthropic의 MCP(Model Context Protocol) 구현.
https://github.com/user-attachments/assets/e68e873f-f87b-4afd-874f-157086e223af
Related MCP server: MSSQL MCP Server
MCP Firebird는 Firebird SQL 데이터베이스 용 Anthropic의 모델 컨텍스트 프로토콜(MCP)을 구현하는 서버입니다. Claude와 같은 대규모 언어 모델(LLM)이 Firebird 데이터베이스의 데이터에 안전하고 통제된 방식으로 접근, 분석 및 조작할 수 있도록 지원합니다.
SQL 쿼리 : Firebird 데이터베이스에서 SQL 쿼리 실행
스키마 분석 : 테이블, 열 및 관계에 대한 자세한 정보를 얻습니다.
데이터베이스 관리 : 백업, 복원 및 검증 작업 수행
성능 분석 : 쿼리 성능을 분석하고 최적화를 제안합니다.
다중 전송 : STDIO 및 SSE(Server-Sent Events) 전송 지원
Claude 통합 : Claude Desktop 및 기타 MCP 클라이언트와 원활하게 작동합니다.
VSCode 통합 : Visual Studio Code에서 GitHub Copilot과 함께 작동
보안 : SQL 쿼리 검증 및 보안 구성 옵션 포함
지엑스피1
백업/복원 작업을 위해서는 Firebird 클라이언트 도구를 설치해야 합니다. 자세한 내용은 "전체 설치"를 참조하세요.
VSCode와 GitHub Copilot 통합에 대해서는 VSCode 통합을 참조하세요.
Claude Desktop 구성을 편집합니다.
code $env:AppData\Claude\claude_desktop_config.json # Windows
code ~/Library/Application\ Support/Claude/claude_desktop_config.json # macOSMCP Firebird 구성을 추가합니다.
{
"mcpServers": {
"mcp-firebird": {
"command": "npx",
"args": [
"mcp-firebird",
"--host",
"localhost",
"--port",
"3050",
"--database",
"C:\\path\\to\\database.fdb",
"--user",
"SYSDBA",
"--password",
"masterkey"
],
"type": "stdio"
}
}
}Claude Desktop을 다시 시작하세요
# Start with SSE transport
npx mcp-firebird --transport-type sse --sse-port 3003 --database /path/to/database.fdbSmithery 를 통해 Claude Desktop에 MCP Firebird를 자동으로 설치하려면:
npx -y @smithery/cli install @PuroDelphi/mcpFirebird --client claude더 자세한 내용은 다음 문서를 확인하세요.
MCP Firebird가 여러분의 업무나 프로젝트에 유용하다고 생각되시면 기부를 통해 개발을 지원해 주세요. 여러분의 기여는 이 도구를 유지하고 개선하는 데 도움이 됩니다.
GitHub 스폰서 : 스폰서 @PuroDelphi
PayPal : PayPal을 통해 기부하세요
이 프로젝트를 지원하는 또 다른 좋은 방법은 Asistentes Autónomos를 통해 AI 에이전트를 고용하는 것입니다. 저희는 다양한 비즈니스 요구에 맞춰 전문화된 AI 비서를 제공하여 업무를 자동화하고 생산성을 향상하는 데 도움을 드립니다.
⭐ 기부자, 스폰서, 그리고 고객은 문제 해결, 기능 요청 및 구현 지침에 대한 우선적인 지원과 도움을 받습니다 . 저희는 모든 사용자를 돕기 위해 최선을 다하지만, 프로젝트에 재정적으로 지원해 주시는 분들께는 더 빠른 응답과 전담 지원을 받으실 수 있습니다.
귀하의 지원에 깊이 감사드립니다. 귀하의 지원은 MCP Firebird의 지속적인 개발을 보장하는 데 큰 도움이 됩니다!
이 프로젝트는 MIT 라이선스에 따라 라이선스가 부여되었습니다. 자세한 내용은 라이선스 파일을 참조하세요.
Suggests indexes for simple single-table SELECT predicates/order terms after checking existing catalog indexes, including primary keys. Heuristic advice only; never executes the query or applies index DDL.
Inputs: sql (required). Returns: error, result, success
Analyzes the performance of a SQL query by executing it multiple times and measuring execution time
Inputs: sql (required), params, iterations. Returns: error, result, success
Analyzes statistical information about a table including row count, column statistics, and data distribution.
Inputs: tableName (required). Returns: error, result, success
Gets the detailed schema of multiple tables in parallel for improved performance.
Inputs: tableNames (required), maxConcurrent. Returns: error, result, success
Gets detailed information about a specific function, including its source code (for PSQL functions)
Inputs: functionName (required). Returns: error, result, success
Gets detailed information about a specific package, including its header and body source
Inputs: packageName (required). Returns: error, result, success
Gets detailed information about a specific stored procedure, including its source code and parameters
Inputs: procedureName (required). Returns: error, result, success
Gets the detailed schema (columns, types, etc.) of a specific table.
Inputs: tableName (required). Returns: error, result, success
Gets detailed information about a specific trigger, including its source code, type, sequence, and status
Inputs: triggerName (required). Returns: error, result, success
Echoes the original message in structured data with markup-encoded display text
Inputs: message (required). Returns: error, result, success
Executes multiple SQL queries in parallel for improved performance.
Inputs: queries (required), maxConcurrent. Returns: error, result, success
Executes a SQL query in the Firebird database. Uses FIRST/ROWS for pagination.
Inputs: sql (required), params. Returns: error, result, success
Retrieves general information about the connected Firebird database.
Inputs: no named parameters. Returns: error, result, success
Prepares an authorized SELECT in a read-only transaction and retrieves its optimizer plan without executing it. Parameters are not bound; unavailable plans return an error, never an execution fallback.
Inputs: sql (required), params. Returns: error, result, success
Gets the stored descriptions for fields of a specific table (if they exist).
Inputs: tableName (required). Returns: error, result, success
Gets information about the Firebird MCP server and its available tools
Inputs: no named parameters. Returns: error, result, success
Returns PRIMARY KEY, FOREIGN KEY, UNIQUE, NOT NULL, and CHECK constraints for a table, including constrained columns and references.
Inputs: tableName (required). Returns: error, result, success
Retrieves data from a specific table with optional filtering, pagination, and ordering.
Inputs: tableName (required), skip, first, filters, +1 more. Returns: error, result, success
Returns all indexes for a table, including ordered columns, uniqueness, direction, and segment count.
Inputs: tableName (required). Returns: error, result, success
Returns only the triggers associated with a table, including source, description, sequence, type, and active state.
Inputs: tableName (required). Returns: error, result, success
Gets detailed information about a specific tool
Inputs: toolName (required). Returns: error, result, success
Lists native events (POST_EVENT) available in database triggers and procedures
Inputs: no named parameters. Returns: error, result, success
Lists all tools available on the MCP server
Inputs: category. Returns: error, result, success
Lists all functions in the database (UDFs and PSQL functions)
Inputs: no named parameters. Returns: error, result, success
Lists all packages in the database (available in Firebird 3.0+)
Inputs: no named parameters. Returns: error, result, success
Lists all stored procedures in the database with input and output parameter information
Inputs: no named parameters. Returns: error, result, success
Lists all user tables in the current Firebird database.
Inputs: no named parameters. Returns: error, result, success
Lists all database triggers with their associated table, trigger type, and status
Inputs: no named parameters. Returns: error, result, success
Register a Firebird POST_EVENT listener; MCP resource subscription is also required for notifications
Inputs: eventName (required). Returns: error, result, success
Reports process health and runtime information; does not probe database connectivity
Inputs: no named parameters. Returns: error, result, success
Release a Firebird event registration owned by this connection or HTTP application
Inputs: eventName (required). Returns: error, result, success
Reports wire-encryption configuration; does not verify negotiated encryption on an active database connection.
Inputs: no named parameters. Returns: error, result, success
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 32 tools
Most tools target clearly distinct resources and actions, especially the table, trigger, procedure, function, package, and query-analysis families. There is some overlap among meta/introspection tools such as get-server-info vs list-available-tools, and between list-triggers, get-table-triggers, and describe-trigger, but descriptions generally make the intended use clear.
Most tool names follow a predictable kebab-case verb_noun pattern such as list-tables, describe-table, get-table-indexes, and execute-query. The main deviations are subscribe_to_event and unsubscribe_from_event using snake_case, plus a few noun-only names like echo, but the overall convention remains readable and mostly consistent.
With 32 tools, the server is well above the typical well-scoped range and includes several niche or overlapping meta tools such as list-available-tools, get-server-info, system-health-check, and batch variants. The breadth reflects a comprehensive Firebird surface, but the count feels heavy rather than tightly curated.
Core Firebird workflows are well covered: tables, schemas, data access, indexes, constraints, triggers, procedures, functions, packages, query execution, plans, performance analysis, events, and database/server metadata. Some object categories such as views, generators/sequences, and privileges lack dedicated list/describe tools, though agents can query system tables via execute-query as a workaround.
Text-to-SQL MCP server: read-only queries on PostgreSQL, MySQL and SQL Server with your real schema
Hosted SQL Server MCP server for Claude, ChatGPT and Cursor, with each action switched per account.
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
Hosted PostgreSQL MCP server for Claude, ChatGPT and Cursor, with each action switched per account.