Deepseek MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| HTTP_PORT | No | HTTP server port (when TRANSPORT=http) | 3000 |
| TRANSPORT | No | Transport mode: stdio or http | stdio |
| MAX_RETRIES | No | Maximum retry count for failed requests | 2 |
| MAX_SESSIONS | No | Maximum number of concurrent sessions | 100 |
| DEFAULT_MODEL | No | Default model for requests | deepseek-chat |
| SHOW_COST_INFO | No | Show cost info in responses | true |
| REQUEST_TIMEOUT | No | Request timeout in milliseconds | 60000 |
| DEEPSEEK_API_KEY | Yes | Your DeepSeek API key | |
| FALLBACK_ENABLED | No | Enable automatic model fallback on errors | true |
| DEEPSEEK_BASE_URL | No | Custom API endpoint | https://api.deepseek.com |
| ENABLE_MULTIMODAL | No | Enable multimodal (image) input support | false |
| MAX_MESSAGE_LENGTH | No | Maximum message content length (characters) | 100000 |
| SESSION_TTL_MINUTES | No | Session time-to-live in minutes | 30 |
| MAX_SESSION_MESSAGES | No | Max messages per session (sliding window) | 200 |
| SKIP_CONNECTION_TEST | No | Skip startup API connection test | false |
| CIRCUIT_BREAKER_THRESHOLD | No | Consecutive failures before circuit opens | 5 |
| CIRCUIT_BREAKER_RESET_TIMEOUT | No | Milliseconds before circuit half-opens | 30000 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": true
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| deepseek_chatB | Chat with DeepSeek models. deepseek-flash (V4.1 Flash, fast and economical) and deepseek-v4-pro (V4 Pro), both 1M context with optional chain-of-thought thinking mode. deepseek-v4-flash, deepseek-chat and deepseek-reasoner are legacy names, still accepted for backward compatibility (they resolve to deepseek-flash); prefer deepseek-flash. Features: multi-turn sessions (session_id), function calling (tools parameter), thinking mode, JSON output mode, multimodal input (when enabled), automatic cost tracking, and model fallback with circuit breaker resilience. |
| deepseek_fimA | Fill-in-the-Middle (FIM) completion with DeepSeek. Provide a prompt (prefix) and an optional suffix; the model completes the text in between. Ideal for code completion and content infilling. Runs in non-thinking mode on the Beta endpoint; output is capped at 4K tokens. The legacy names deepseek-v4-flash, deepseek-chat and deepseek-reasoner are still accepted and resolve to deepseek-flash (FIM has no thinking mode). Includes automatic cost tracking and model fallback with circuit breaker resilience. |
| deepseek_sessionsA | Manage multi-turn conversation sessions. List active sessions, delete a specific session, or clear all sessions. Sessions store conversation history for use with the session_id parameter in deepseek_chat. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| debug_with_reasoning | Debug code issues using DeepSeek R1 reasoning model with step-by-step analysis |
| code_review_deep | Comprehensive code review analyzing quality, security, performance, and best practices |
| research_synthesis | Research a topic and synthesize information into a structured report |
| strategic_planning | Analyze options and create strategic plans with reasoning for each decision |
| explain_like_im_five | Explain complex topics in simple terms using analogies and reasoning |
| mathematical_proof | Prove mathematical statements with rigorous step-by-step reasoning |
| argument_validation | Analyze arguments for logical fallacies and reasoning errors |
| creative_ideation | Generate creative ideas with reasoning for feasibility and value |
| cost_comparison | Compare costs of different LLMs for a task and show savings with DeepSeek |
| pair_programming | Interactive coding assistant that explains reasoning for code decisions |
| function_call_debug | Debug function calling issues with DeepSeek models |
| create_function_schema | Generate JSON Schema for function calling from natural language description |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| models | List of available DeepSeek models with capabilities, context limits, and pricing information |
| config | Current server configuration including base URL, timeouts, session settings, and fallback status. API key is masked for security. |
| usage | Real-time usage statistics including total requests, token consumption, costs, active sessions, and cache hit ratio. Updated on every read. |
TDQS
Scored across 3 tools
deepseek_chat and deepseek_fim both produce model completions, which could be confused, but the descriptions clearly distinguish conversational/multi-turn use from prefix-suffix infilling. deepseek_sessions is clearly separate as a management utility.
All three tools use a consistent deepseek_<noun/verb> snake_case pattern (chat, fim, sessions) with no style mixing.
Three tools is on the thin side, but each covers a distinct capability (chat, FIM, session management) and earns its place for a chat API wrapper. Slight deduction because deepseek_sessions bundles list/delete/clear operations rather than exposing them separately.
Core chat and completion workflows are covered, and sessions can be listed/deleted/cleared, but there's no explicit session creation and no embeddings or model-listing tool, which are typical gaps for an LLM API server. Agents can mostly work around this since sessions appear to auto-create via session_id.