Skip to content

Anonymous Chat Interactions

Operations Summary

Method Endpoint Description
Core CRUD
GET /api/anonymous-chat-interactions/ List Anonymous Chat Interactions
GET /api/anonymous-chat-interactions/{uuid}/ Retrieve
Other Actions
GET /api/anonymous-chat-interactions/budget/ Today's global tenant budget snapshot
GET /api/anonymous-chat-interactions/by-session/{session_id}/ Full transcript for one anonymous session
GET /api/anonymous-chat-interactions/by-user/ Aggregate user list (no slug)
GET /api/anonymous-chat-interactions/by-user/{user_slug}/ All sessions for one pseudonymous user
GET /api/anonymous-chat-interactions/conversations/ Conversations grouped by session
GET /api/anonymous-chat-interactions/kpi/ Aggregate KPI roll-up

Core CRUD

List Anonymous Chat Interactions

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.models.anonymous_chat_interaction_o_enum import AnonymousChatInteractionOEnum # (1)
from waldur_api_client.models.injection_severity_enum import InjectionSeverityEnum # (2)
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_list # (3)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_list.sync(client=client)

for item in response:
    print(item)
  1. Model Source: AnonymousChatInteractionOEnum
  2. Model Source: InjectionSeverityEnum
  3. API Source: anonymous_chat_interactions_list
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
import { anonymousChatInteractionsList } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsList({
  auth: "Token YOUR_API_TOKEN"
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Description
created_after string (date)
created_before string (date)
has_feedback boolean
is_flagged boolean
o array Ordering
page integer A page number within the paginated result set.
page_size integer Number of results to return per page.
query string
session_id string
severity string Enum: none, low, medium, high, critical
user_slug string

200 -

The response body is an array of objects, where each object has the following structure:

Field Type Description
uuid string (uuid)
user_slug string
user_input string
assistant_blocks object (free-form)
click_count integer Offering click-throughs on this interaction. Populated by the by-session transcript, which annotates it; 0 elsewhere.
input_tokens integer
output_tokens integer
offering_uuids object (free-form)
result_count integer
is_flagged boolean
severity string
injection_categories object (free-form)
pii_categories object (free-form)
action_taken string
warning string
ip_address any An IPv4 or IPv6 address.
session_id string
last_active_at string (date-time)
created string (date-time)
feedback any

Retrieve

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/a1b2c3d4-e5f6-7890-abcd-ef1234567890/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_retrieve # (1)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_retrieve.sync(
    uuid="a1b2c3d4-e5f6-7890-abcd-ef1234567890",
    client=client
)

print(response)
  1. API Source: anonymous_chat_interactions_retrieve
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
import { anonymousChatInteractionsRetrieve } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsRetrieve({
  auth: "Token YOUR_API_TOKEN",
  path: {
    "uuid": "a1b2c3d4-e5f6-7890-abcd-ef1234567890"
  }
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Required
uuid string (uuid) ✓

200 -

Field Type Description
uuid string (uuid)
user_slug string
user_input string
assistant_blocks object (free-form)
click_count integer Offering click-throughs on this interaction. Populated by the by-session transcript, which annotates it; 0 elsewhere.
input_tokens integer
output_tokens integer
offering_uuids object (free-form)
result_count integer
is_flagged boolean
severity string
injection_categories object (free-form)
pii_categories object (free-form)
action_taken string
warning string
ip_address any An IPv4 or IPv6 address.
session_id string
last_active_at string (date-time)
created string (date-time)
feedback any

Other Actions

Today's global tenant budget snapshot

Returns the site-wide token + request usage accumulated since 00:00 UTC today and the configured daily caps. Powers the budget gauges card on the staff analytics dashboard.

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/budget/ \
  Authorization:"Token YOUR_API_TOKEN"

1
2
3
4
5
6
7
8
9
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_budget_retrieve # (1)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_budget_retrieve.sync(client=client)

print(response)
  1. API Source: anonymous_chat_interactions_budget_retrieve
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
import { anonymousChatInteractionsBudgetRetrieve } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsBudgetRetrieve({
  auth: "Token YOUR_API_TOKEN"
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}

200 -

Field Type
tokens_today integer
tokens_limit integer
resets_at string (date-time)

Full transcript for one anonymous session

Returns the ordered list of interactions belonging to the given session_id. Use this to read a conversation as a transcript.

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/by-session/string-value/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.models.anonymous_chat_interaction_o_enum import AnonymousChatInteractionOEnum # (1)
from waldur_api_client.models.injection_severity_enum import InjectionSeverityEnum # (2)
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_by_session_list # (3)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_by_session_list.sync(
    session_id="string-value",
    client=client
)

for item in response:
    print(item)
  1. Model Source: AnonymousChatInteractionOEnum
  2. Model Source: InjectionSeverityEnum
  3. API Source: anonymous_chat_interactions_by_session_list
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
import { anonymousChatInteractionsBySessionList } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsBySessionList({
  auth: "Token YOUR_API_TOKEN",
  path: {
    "session_id": "string-value"
  }
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Required
session_id string ✓
Name Type Description
created_after string (date)
created_before string (date)
has_feedback boolean
is_flagged boolean
o array Ordering
page integer A page number within the paginated result set.
page_size integer Number of results to return per page.
query string
session_id string
severity string Enum: none, low, medium, high, critical
user_slug string

200 -

The response body is an array of objects, where each object has the following structure:

Field Type Description
uuid string (uuid)
user_slug string
user_input string
assistant_blocks object (free-form)
click_count integer Offering click-throughs on this interaction. Populated by the by-session transcript, which annotates it; 0 elsewhere.
input_tokens integer
output_tokens integer
offering_uuids object (free-form)
result_count integer
is_flagged boolean
severity string
injection_categories object (free-form)
pii_categories object (free-form)
action_taken string
warning string
ip_address any An IPv4 or IPv6 address.
session_id string
last_active_at string (date-time)
created string (date-time)
feedback any

Aggregate user list (no slug)

Returns one row per user_slug with aggregate counters. Powers the staff Users page in the admin analytics.

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/by-user/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.models.anonymous_chat_interaction_o_enum import AnonymousChatInteractionOEnum # (1)
from waldur_api_client.models.injection_severity_enum import InjectionSeverityEnum # (2)
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_by_user_aggregate # (3)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_by_user_aggregate.sync(client=client)

for item in response:
    print(item)
  1. Model Source: AnonymousChatInteractionOEnum
  2. Model Source: InjectionSeverityEnum
  3. API Source: anonymous_chat_interactions_by_user_aggregate
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
import { anonymousChatInteractionsByUserAggregate } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsByUserAggregate({
  auth: "Token YOUR_API_TOKEN"
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Description
created_after string (date)
created_before string (date)
has_feedback boolean
is_flagged boolean
o array Ordering
page integer A page number within the paginated result set.
page_size integer Number of results to return per page.
query string
session_id string
severity string Enum: none, low, medium, high, critical
user_slug string

200 -

The response body is an array of objects, where each object has the following structure:

Field Type
user_slug string
last_seen string (date-time)
total_interactions integer
session_count integer
positive_feedback integer
negative_feedback integer
no_feedback integer
injection_strikes integer

All sessions for one pseudonymous user

Returns interactions sharing a user_slug (Scrypt of originating IP) — across however many sessions that anon user opened, ordered chronologically.

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/by-user/string-value/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.models.anonymous_chat_interaction_o_enum import AnonymousChatInteractionOEnum # (1)
from waldur_api_client.models.injection_severity_enum import InjectionSeverityEnum # (2)
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_by_user_list # (3)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_by_user_list.sync(
    user_slug="string-value",
    client=client
)

for item in response:
    print(item)
  1. Model Source: AnonymousChatInteractionOEnum
  2. Model Source: InjectionSeverityEnum
  3. API Source: anonymous_chat_interactions_by_user_list
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
import { anonymousChatInteractionsByUserList } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsByUserList({
  auth: "Token YOUR_API_TOKEN",
  path: {
    "user_slug": "string-value"
  }
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Required
user_slug string ✓
Name Type Description
created_after string (date)
created_before string (date)
has_feedback boolean
is_flagged boolean
o array Ordering
page integer A page number within the paginated result set.
page_size integer Number of results to return per page.
query string
session_id string
severity string Enum: none, low, medium, high, critical
user_slug string

200 -

The response body is an array of objects, where each object has the following structure:

Field Type Description
uuid string (uuid)
user_slug string
user_input string
assistant_blocks object (free-form)
click_count integer Offering click-throughs on this interaction. Populated by the by-session transcript, which annotates it; 0 elsewhere.
input_tokens integer
output_tokens integer
offering_uuids object (free-form)
result_count integer
is_flagged boolean
severity string
injection_categories object (free-form)
pii_categories object (free-form)
action_taken string
warning string
ip_address any An IPv4 or IPv6 address.
session_id string
last_active_at string (date-time)
created string (date-time)
feedback any

Conversations grouped by session

One row per anonymous conversation (session_id) with aggregate counters, mirroring the authenticated threads table. Honours the same filters as the list endpoint. Read a conversation's transcript via by-session/{session_id}.

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/conversations/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.models.anonymous_chat_interaction_o_enum import AnonymousChatInteractionOEnum # (1)
from waldur_api_client.models.injection_severity_enum import InjectionSeverityEnum # (2)
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_conversations_list # (3)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_conversations_list.sync(client=client)

for item in response:
    print(item)
  1. Model Source: AnonymousChatInteractionOEnum
  2. Model Source: InjectionSeverityEnum
  3. API Source: anonymous_chat_interactions_conversations_list
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
import { anonymousChatInteractionsConversationsList } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsConversationsList({
  auth: "Token YOUR_API_TOKEN"
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Description
created_after string (date)
created_before string (date)
has_feedback boolean
input_tokens_max number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
input_tokens_min number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
is_flagged boolean
is_reviewed boolean Whether the nightly LLM judge has scored the conversation.
last_active_after string (date) Bound on the conversation's most recent turn. Inclusive of the boundary day.
last_active_before string (date) Bound on the conversation's most recent turn. Inclusive of the boundary day.
o array Ordering
output_tokens_max number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
output_tokens_min number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
page integer A page number within the paginated result set.
page_size integer Number of results to return per page.
query string
session_id string
severity string Enum: none, low, medium, high, critical
total_tokens_max number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
total_tokens_min number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
user_slug string

200 -

The response body is an array of objects, where each object has the following structure:

Field Type Description
session_id string
user_slug string
message_count integer
is_flagged boolean
max_severity string
has_feedback boolean
offerings_shown integer
offerings_clicked integer Click-throughs on recommended offerings; repeat clicks count separately.
models_used string Comma-separated distinct LLM models across the conversation. More than one when AI_ASSISTANT_MODEL was switched partway through; blank for conversations predating model tracking.
is_reviewed boolean True once the nightly LLM judge has scored this conversation. One verdict per conversation, recorded on its last turn; a conversation is never re-judged.
input_tokens integer Prompt tokens summed over the conversation; 0 for turns predating token tracking.
output_tokens integer Completion tokens summed over the conversation.
total_tokens integer input_tokens + output_tokens. Excludes LLM judge spend, which runs on its own budget.
started string (date-time)
last_active string (date-time)

Aggregate KPI roll-up

Returns aggregate counters and rates for the anonymous chat flow. Filters are honoured (date range etc.) so the same parameters work as on the list endpoint.

1
2
3
4
http \
  GET \
  https://api.example.com/api/anonymous-chat-interactions/kpi/ \
  Authorization:"Token YOUR_API_TOKEN"

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from waldur_api_client.client import AuthenticatedClient
from waldur_api_client.models.anonymous_chat_interaction_o_enum import AnonymousChatInteractionOEnum # (1)
from waldur_api_client.models.injection_severity_enum import InjectionSeverityEnum # (2)
from waldur_api_client.api.anonymous_chat_interactions import anonymous_chat_interactions_kpi_retrieve # (3)

client = AuthenticatedClient(
    base_url="https://api.example.com", token="YOUR_API_TOKEN"
)
response = anonymous_chat_interactions_kpi_retrieve.sync(client=client)

print(response)
  1. Model Source: AnonymousChatInteractionOEnum
  2. Model Source: InjectionSeverityEnum
  3. API Source: anonymous_chat_interactions_kpi_retrieve
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
import { anonymousChatInteractionsKpiRetrieve } from 'waldur-js-client';

try {
  const response = await anonymousChatInteractionsKpiRetrieve({
  auth: "Token YOUR_API_TOKEN"
});
  console.log('Success:', response);
} catch (error) {
  console.error('Error:', error);
}
Name Type Description
created_after string (date)
created_before string (date)
has_feedback boolean
input_tokens_max number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
input_tokens_min number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
is_flagged boolean
is_reviewed boolean Whether the nightly LLM judge has scored the conversation.
last_active_after string (date) Bound on the conversation's most recent turn. Inclusive of the boundary day.
last_active_before string (date) Bound on the conversation's most recent turn. Inclusive of the boundary day.
o array Ordering
output_tokens_max number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
output_tokens_min number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
query string
session_id string
severity string Enum: none, low, medium, high, critical
total_tokens_max number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
total_tokens_min number Conversation-level bound: selects whole conversations by their summed spend, never individual turns.
user_slug string

200 -

Field Type Description
interactions_total integer
sessions_total integer
unique_users integer Distinct user_slug values in the window — proxy for active anonymous users.
flagged_total integer
feedback_positive integer
feedback_negative integer
satisfaction_rate number (double) positive / (positive + negative); null when no human feedback.
clicks_total integer
click_through_rate number (double) clicks / interactions; null when no interactions.
input_tokens_total integer Prompt tokens summed over the filtered turns. Turns recorded before per-interaction token capture contribute nothing, so this understates spend on historical data.
output_tokens_total integer Completion tokens summed over the filtered turns.
reviewed_total integer Threads carrying a judge verdict. Always present, unlike the review rates below — zero here is the signal that the nightly pass is off or stalled, so consumers can keep showing the row.
review_input_tokens_total integer Prompt tokens spent by the LLM judge. Tracked apart from input_tokens_total because review runs on its own budget.
review_output_tokens_total integer Completion tokens spent by the LLM judge.
avg_llm_resolution_score number (double) Mean of llm_resolution_score across reviewed sessions (1-5).
llm_intent_distribution object (free-form) Counts keyed by llm_intent_category.
hallucination_rate number (double) Share of reviewed sessions flagged as hallucinating.
review_coverage number (double) Reviewed sessions / total reviewable sessions. Operations health signal — drops below ~90% if the review budget is too tight or the task is failing.
daily_volume array of objects Per-day query counts across the filter window.
daily_volume.date string (date)
daily_volume.count integer
severity_by_day any Stacked-bar input. Shape: {labels: [iso-date], series: {NONE: [...], LOW: [...], MEDIUM: [...], HIGH: [...], CRITICAL: [...]}}