List, create, configure, and manage Charms — everything except running them.
|
|
https://api.charmiq.ai/mcp/agentcharmiq://charm/<id> document it was created against and that binding doesn't move.list_agents | |
list_agent_instructions | |
get_agent | |
get_agent_instructions | |
set_agent_instructions | |
describe_model_config | |
create_agent | |
config_agent | |
copy_agent | |
delete_agent |
create_agent with systemPrompt (inline text) or no arguments at all (empty Charm). Then config_agent to set the model, and optionally set_agent_instructions to refine the prompt.list_agent_instructions to find a System Prompt you want to bind to. Call create_agent with systemPromptUri pointing at it. Then config_agent for the model.copy_agent — same instructions, independent config. Then config_agent or set_agent_instructions to customize.set_agent_instructions (which affects every Charm bound to it) or create a new Charm.config_agent updates the model, temperature, tools, MCP servers and other LLM settings — and optionally renames the Charm. Before calling it, use describe_model_config to learn what fields are valid for the provider and model you're targeting. The response includes:describe_model_config first. Models change. Capability surfaces change. Read what's currently valid; don't hard-code from documentation.set_agent_instructions is a full overwrite. If multiple Charms share the System Prompt, you're editing all of them.copy_agent for experiments. Cheaper than rebuilding a Charm from scratch when you want to try a different model or temperature.charmiq://charm/<id> document. A Charm is a (systemPromptId, llmConfig) pair. The same System Prompt can have many Charms; the same Charm has exactly one System Prompt.charmiq://charm/… documents that back Charm Instructions.