REST API — persona

Backfill embeddings for pending memories

Create a global (persona-scope) memory

Semantic recall of a persona’s memories (KNN)

List models

Create models

Get models by id

Update models

Delete models

List personas

Create personas

Get personas by id

Update personas

Delete personas

List/search LLM provider templates (public; ?q= substring on name/kind, ?providerKind= exact, paginated)

Create a provider template with its models in one call (admin)

One provider template with its models (public)

Update a provider template; optional full model-set sync (admin)

Delete a provider template and its models (admin)

Probe the live provider with a caller-supplied key — never stored — to verify kind/baseUrl speak the claimed contract (admin)

List service_templates

Create service_templates

Get service_templates by id

Update service_templates

Delete service_templates

List the caller's LLM provider connections

Create an LLM provider connection

Get an LLM provider connection by id

Update an LLM provider connection

Delete an LLM provider connection

Test the connection — probe the provider, persist status + cache the catalog

Live provider model catalog — fetched fresh, not persisted

Re-fetch + persist the provider catalog (probes the provider, like connect)

List tool_templates

Create tool_templates

Get tool_templates by id

Update tool_templates

Delete tool_templates