LoomConfiguration

Long-term memory for an interaction’s agents (loom = { ... } in its definition file): whose loom each agent wakes up with, and how memories are recalled and made. Fields left null take their defaults; the fields whose names start with effective give the values in use.

Kind: an object, used in requests and responses. In a request it is called LoomConfigurationInput.

In a request

Field	Type	Required	Description
enabled	boolean		Whether memory is on. False keeps the settings but leaves memory off.
scope	LoomScope| null		Whose loom an agent wakes up with, and so whether it remembers beyond one run. Null: Process.
scopeInput	string | null		For the Input scope: the run input whose value picks the loom, such as "userName".
shared	boolean | null		True: all the interaction’s agents share one loom. Null or false: each agent has its own.
recall	LoomRecallMode| null		How remembered threads come back into a conversation. Null: Suggest.
weave	LoomWeaveMode| null		How new threads are made from a conversation. Null: Suggest.
tapestryTokens	integer | null		How long the “what you know” summary (the tapestry) an agent is given at the start of a conversation may be, in tokens, from 200 to 20,000. Null: 1,500.
subjectInputs	string[] | null		Run inputs that name who the agents are working with, such as "userName", so memories are about the right person (as well as the people who send messages).
nightWeaving	boolean | null		Whether looms this interaction creates are tidied up by night weaving once a conversation goes quiet.
quietMinutes	integer | null		For looms this interaction creates: minutes without activity before night weaving starts.
weaverModel	string | null		The catalogue model that reads conversations to make memories and to night-weave. Null: the agent’s own model.
vetModel	string | null		A lightweight catalogue model that checks what the loom is about to offer an agent mid-conversation and drops what clearly does not belong, before the agent sees it. It can only take things away; if it fails or is slow, everything is offered as usual. A run can override it per agent. Null: nothing is checked.
vetRecall	boolean | null		True: the vetModel also checks what an agent’s own searches of its memory bring back. Null or false: a search the agent chose is its own to judge.
seedFromLoomId	uuid | null		The id of a loom every loom this interaction creates starts from: a new person’s loom starts knowing what that loom knows (its shared facts, how-tos and relationships, each marked as inherited).
readFromLoomIds	uuid[] | null		The ids of looms every agent of this interaction may read but never change: recall and suggestions reach into them, each thread labelled with where it lives.

{
  "enabled": false,
  "scope": "Process",
  "scopeInput": "...",
  "shared": false,
  "recall": "Off",
  "weave": "Manual",
  "tapestryTokens": 0,
  "subjectInputs": [
    "..."
  ],
  "nightWeaving": false,
  "quietMinutes": 0,
  "weaverModel": "...",
  "vetModel": "...",
  "vetRecall": false,
  "seedFromLoomId": "3f2b8c1e-5d4a-4e6b-9c7d-2a1b0c9d8e7f",
  "readFromLoomIds": [
    "3f2b8c1e-5d4a-4e6b-9c7d-2a1b0c9d8e7f"
  ]
}

In a response

Every field is present in a response; empty ones are null.

Field	Type	Description
enabled	boolean	Whether memory is on. False keeps the settings but leaves memory off.
scope	LoomScope| null	Whose loom an agent wakes up with, and so whether it remembers beyond one run. Null: Process.
scopeInput	string | null	For the Input scope: the run input whose value picks the loom, such as "userName".
shared	boolean | null	True: all the interaction’s agents share one loom. Null or false: each agent has its own.
recall	LoomRecallMode| null	How remembered threads come back into a conversation. Null: Suggest.
weave	LoomWeaveMode| null	How new threads are made from a conversation. Null: Suggest.
tapestryTokens	integer | null	How long the “what you know” summary (the tapestry) an agent is given at the start of a conversation may be, in tokens, from 200 to 20,000. Null: 1,500.
subjectInputs	string[] | null	Run inputs that name who the agents are working with, such as "userName", so memories are about the right person (as well as the people who send messages).
nightWeaving	boolean | null	Whether looms this interaction creates are tidied up by night weaving once a conversation goes quiet.
quietMinutes	integer | null	For looms this interaction creates: minutes without activity before night weaving starts.
weaverModel	string | null	The catalogue model that reads conversations to make memories and to night-weave. Null: the agent’s own model.
vetModel	string | null	A lightweight catalogue model that checks what the loom is about to offer an agent mid-conversation and drops what clearly does not belong, before the agent sees it. It can only take things away; if it fails or is slow, everything is offered as usual. A run can override it per agent. Null: nothing is checked.
vetRecall	boolean | null	True: the vetModel also checks what an agent’s own searches of its memory bring back. Null or false: a search the agent chose is its own to judge.
seedFromLoomId	uuid | null	The id of a loom every loom this interaction creates starts from: a new person’s loom starts knowing what that loom knows (its shared facts, how-tos and relationships, each marked as inherited).
readFromLoomIds	uuid[] | null	The ids of looms every agent of this interaction may read but never change: recall and suggestions reach into them, each thread labelled with where it lives.
effectiveScope	LoomScope	The scope in use: scope, or Process when that is null. Set by Maitento.
effectiveShared	boolean	Whether the agents share one loom: shared, or false when that is null. Set by Maitento.
effectiveRecall	LoomRecallMode	The recall mode in use: recall, or Suggest when that is null. Set by Maitento.
effectiveWeave	LoomWeaveMode	The weave mode in use: weave, or Suggest when that is null. Set by Maitento.
effectiveTapestryTokens	integer	The tapestry length in use, in tokens: tapestryTokens (or 1,500), kept between 200 and 20,000. Set by Maitento.
effectiveSubjectInputs	string[]	The subject inputs in use: subjectInputs, or none when that is null. Set by Maitento.
effectiveReadFromLoomIds	uuid[]	The read-only looms in use: readFromLoomIds, or none when that is null. Set by Maitento.
vets	boolean	Whether a vetModel is set, so what the loom offers is checked. Set by Maitento.
effectiveVetRecall	boolean	Whether the agents’ own searches are checked too: vetRecall, and only when there is a vetModel. Set by Maitento.

{
  "enabled": false,
  "scope": "Process",
  "scopeInput": "...",
  "shared": false,
  "recall": "Off",
  "weave": "Manual",
  "tapestryTokens": 0,
  "subjectInputs": [
    "..."
  ],
  "nightWeaving": false,
  "quietMinutes": 0,
  "weaverModel": "...",
  "vetModel": "...",
  "vetRecall": false,
  "seedFromLoomId": "3f2b8c1e-5d4a-4e6b-9c7d-2a1b0c9d8e7f",
  "readFromLoomIds": [
    "3f2b8c1e-5d4a-4e6b-9c7d-2a1b0c9d8e7f"
  ],
  "effectiveScope": "Process",
  "effectiveShared": false,
  "effectiveRecall": "Off",
  "effectiveWeave": "Manual",
  "effectiveTapestryTokens": 0,
  "effectiveSubjectInputs": [
    "..."
  ],
  "effectiveReadFromLoomIds": [
    "3f2b8c1e-5d4a-4e6b-9c7d-2a1b0c9d8e7f"
  ],
  "vets": false,
  "effectiveVetRecall": false
}

Where it’s used

• Inside ManagedInteractionCreateRequest
• Inside ManagedInteractionUpdateRequest
• Inside ManagedInteractionVersion
• Inside OneShotInteractionCreateRequest
• Inside OneShotInteractionUpdateRequest
• Inside OneShotInteractionVersion
• Inside RoundRobinInteractionCreateRequest
• Inside RoundRobinInteractionUpdateRequest
• Inside RoundRobinInteractionVersion