About Llauncher
Llauncher is an MCP server in the Agents category: llama-server model endpoint manager with JiT swapping with human UI, MCP and pi extension support.
Install
git clone https://github.com/shanevcantwell/llauncherThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Tools (17)
add_modelAdd a new model configuration to the store
Parameters not published.
cancel_serverCancel an in-flight start/swap on a port (ADR-LLNCH-014)
Parameters not published.
delete_modelDelete a model configuration (refuses if running; ADR-LLNCH-008 §4.1)
Parameters not published.
get_model_configGet full configuration details for a specific model
Parameters not published.
get_server_logsFetch recent log lines from a running server
Parameters not published.
list_modelsList all configured models with current status (running/stopped)
Parameters not published.
list_orphansList unmanaged `llama-server` processes on the local node (ADR-LLNCH-015)
Parameters not published.
read_auditRead recent audit-log entries on this node (ADR-LLNCH-008, #64)
Parameters not published.
server_metricsLive inference telemetry for a running server — safe tier, no prompt text (ADR-LLNCH-019)
Parameters not published.
server_slotsPer-slot detail including prompt text — sensitive tier, granted separately from `server_metrics` (ADR-LLNCH-019)
Parameters not published.
Show all 17 tools
server_statusGet status summary of all running servers
Parameters not published.
start_serverStart a llama-server instance on a given port (`model_name` + `port` required; ADR-LLNCH-010)
Parameters not published.
stop_serverStop a running server by port number
Parameters not published.
swap_serverAtomically swap models on a port with rollback guarantee (ADR-LLNCH-011)
Parameters not published.
update_model_configUpdate an existing model's configuration
Parameters not published.
validate_configValidate a configuration without applying it
Parameters not published.
validate_modelsRead-only weight-file validation (existence, GGUF magic, advisory VRAM/lockfile) (#475, ADR-LLNCH-027)
Parameters not published.
Ask AI
Ask AI about Llauncher
Powered by Claude · Grounded in docs
Security checks
6 checks have not been run yet.