sglang — Terradev MCP Tool
sglang
(sglang) is one of 245 tools on the
Terradev
MCP server. Connect the server and your client discovers it on the handshake.
by client
How to call sglang from your client
- Claude Code Terradev sglang run in your project directory
- Claude Desktop Terradev sglang ~/Library/Application Support/Claude/claude_desktop_config.json
- Cursor Terradev sglang ~/.cursor/mcp.json
- VS Code Terradev sglang .vscode/mcp.json
- Zed Terradev sglang ~/.config/zed/settings.json
- Windsurf Terradev sglang ~/.codeium/windsurf/mcp_config.json
- Cline Terradev sglang ~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
- Gemini CLI Terradev sglang ~/.gemini/settings.json
- Grok Terradev sglang .mcp.json (in your project root)
- ChatGPT Terradev sglang Settings → Connectors → Advanced → Developer mode
- Claude.ai Terradev sglang Settings → Connectors → Add custom connector
- LangChain Terradev sglang pip install langchain-mcp-adapters
Fastest route
claude mcp add --transport sse terradev https://terradev-mcp.terradev.cloud/sseOther tools on this server
- provision_gpu
- preflight_report
- preflight_gpu_check
- preflight_network_check
- local_scan
- status
- manage_instance
- analytics
- optimize
- setup_provider
- configure_provider
- train
- preflight
- stage
- up
- rollback
- manifests
- run_workflow
- active_context
- governance_request_consent
- governance_record_consent
- governance_evaluate_opa
- governance_move_data
- governance_movement_history
- governance_compliance_report
- inferx_deploy
- inferx_status
- inferx_list
- inferx_optimize
- inferx_configure
- inferx_delete
- inferx_usage
- inferx_quote
- infer_route
- infer_route_disagg
- infer_failover
- gpu_topology
- vllm_start
- vllm_stop
- vllm_inference
- vllm_info
- vllm_sleep
- vllm_wake
- sglang_start
- sglang_stop
- sglang_inference
- ollama_list
- ollama_pull
- ollama_generate
- ollama_chat
- ollama_model_info
- vllm_auto_optimize
- vllm_analyze
- vllm_benchmark
- ollama_ps
- langfuse_configure
- langfuse_test
- langfuse_traces
- langfuse_trace
- langfuse_scores
- langfuse_score
- langfuse_datasets
- langfuse_export_training_data
- langfuse_quality
- langfuse_otel_env
- langfuse_k8s
- k8s_create
- k8s_list
- k8s_info
- k8s_destroy
- helm_generate
- kserve_generate_yaml
- kserve_list
- kserve_status
- k8s_gpu_operator_install
- k8s_device_plugin
- k8s_mig_configure
- k8s_time_slicing
- hf_space_deploy
- hf_space_status
- ray_status
- ray_start
- ray_stop
- ray_submit_job
- ray_list_jobs
- ray_wide_ep_deploy
- ray_disagg_pd_deploy
- ray_parallelism_strategy
- wandb_list_projects
- wandb_list_runs
- wandb_run_details
- mlflow_list_experiments
- mlflow_log_run
- mlflow_register_model
- dvc_status
- dvc_diff
- dvc_stage_checkpoint
- dvc_push
- hf_list_models
- hf_list_datasets
- hf_model_info
- hf_create_endpoint
- hf_list_endpoints
- hf_endpoint_info
- hf_delete_endpoint
- hf_endpoint_infer
- hf_smart_template
- hf_hardware_recommend
- hf_hardware_compare
- langchain_create_workflow
- langchain_create_sglang_pipeline
- langgraph_create_workflow
- langgraph_orchestrator_worker
- langgraph_evaluation_workflow
- langgraph_workflow_status
- wandb_create_dashboard
- wandb_create_terradev_dashboard
- wandb_create_report
- wandb_create_terradev_report
- wandb_setup_alerts
- wandb_create_terradev_alerts
- wandb_dashboard_status
- phoenix_test
- phoenix_projects
- phoenix_spans
- phoenix_trace
- phoenix_otel_env
- phoenix_snippet
- phoenix_k8s
- guardrails_test
- guardrails_chat
- guardrails_generate_config
- guardrails_k8s
- qdrant_test
- qdrant_collections
- qdrant_create_collection
- qdrant_info
- qdrant_count
- qdrant_k8s
- deepeval_run
- deepeval_metrics
- deepeval_evaluate
- deepeval_init
- egress_cheapest_route
- egress_optimize_staging
- orchestrator_start
- orchestrator_register
- orchestrator_load
- orchestrator_evict
- orchestrator_status
- orchestrator_infer
- warm_pool_start
- warm_pool_status
- price_intel
- cost_analyze
- cost_optimize_recommend
- cost_simulate
- price_trends
- price_spot_risk
- moe_deploy
- train_status
- train_monitor
- checkpoint_list
- checkpoint_save
- train_stop
- train_resume
- checkpoint_restore
- checkpoint_promote
- checkpoint_delete
- lora_list
- lora_add
- lora_remove
- training_config_generate
- training_launch_distributed
- train_snapshot
- train_detect_stragglers
- agent_agentic_serving_configure
- agent_agentic_serving_helm_values
- agent_agentic_serving_k8s
- agent_agentic_serving_launch_args
- agent_agentic_serving_lmcache_env
- agent_agentic_serving_show_config
- agent_cost
- agent_deploy
- agent_langchain_create_langgraph
- agent_langchain_create_pipeline
- agent_langchain_create_workflow
- agent_langchain_test
- agent_langgraph_create_workflow
- agent_langgraph_deploy
- agent_langgraph_status
- agent_langgraph_test
- agent_letta_chat
- agent_letta_create
- agent_letta_delete
- agent_letta_list
- agent_letta_remember
- agent_letta_status
- agent_mem0_add
- agent_mem0_configure
- agent_mem0_delete
- agent_mem0_forget
- agent_mem0_get
- agent_mem0_list
- agent_mem0_search
- agent_mem0_test
- agent_mem0_update
- agent_list
- agent_plan
- agent_scale
- agent_skill_attach
- agent_skill_init
- agent_status
- agent_teardown
- agent_vector_db_down
- agent_vector_db_up
- ml_vllm_lora_link
- ml_vllm_lora_list
- ml_vllm_lora_load
- ml_vllm_lora_sync
- ml_vllm_lora_unload
- database_weaviate_create_collection
- database_weaviate_delete_collection
- database_weaviate_hybrid_search
- database_weaviate_insert
- database_weaviate_list_collections
- database_weaviate_query
- database_weaviate_up
- train_unsloth_run
- train_unsloth_start
- train_unsloth_stop
- create_sqlite_connection
- create_postgresql_connection
- query_database
- upsert_database
- get_database_connection
- secrets_set
- secrets_get
- secrets_list
- secrets_remove
- secrets_sync
- secrets_verify
- secrets_env
- secrets_run
This list was read from the server itself, by connecting to it and calling tools/list on 24 September 2026. It is what
the server actually exposes, not what its listing claims.
Mutating and Read-only are read off each tool's name, not its schema — a hint, not a guarantee. The registry stores tool names only; connect the server for its live schemas.