On this page
- Overview
- List Agent Runs
- Query Parameters
- Example Request
- Response (200 OK)
- Get Agent Run Stats
- Query Parameters
- Example Request
- Response (200 OK)
- Quality: RQS Metrics
- Query Parameters
- Response (200 OK)
- Quality: User Feedback
- Query Parameters
- Response (200 OK)
- Quality: Entropy (Data Quality)
- Query Parameters
- Response (200 OK)
- Admin Portal UI
- Conversations View
- Quality View
Admin
Agent run history, quality metrics, conversation inspector, and usage analytics.
Overview
The Admin API provides observability endpoints for inspecting agent tool execution history, quality metrics, conversation forensics, and usage analytics. The admin portal is accessible at admin.praxiomai.xyz and provides a full UI with conversation inspector, quality dashboards, and entropy monitoring.
These endpoints require workspace membership and are rate-limited to 60 requests per minute.
List Agent Runs
GET /api/admin/agent-runs
List agent runs for a workspace with optional filters. Provides a paginated, filterable view of all agent tool executions (synthesis, recommendation, drafting, chat workflows).
Query Parameters
| Name | Type | Required | Description |
|---|---|---|---|
workspace_id | UUID | Yes | Target workspace |
agent_type | string | No | Filter by type: synthesis, recommendation, drafting, chat |
status | string | No | Filter by status: success, failed, partial |
tool_name | string | No | Filter by specific tool name |
limit | integer | No | Max records per page (default: 50, max: 100) |
offset | integer | No | Pagination offset (default: 0) |
Example Request
curl "https://api.praxiomai.xyz/api/admin/agent-runs?workspace_id=550e8400-...&agent_type=synthesis&limit=10" \
-H "Authorization: Bearer $TOKEN"
import requests
resp = requests.get(
"https://api.praxiomai.xyz/api/admin/agent-runs",
headers={"Authorization": f"Bearer {token}"},
params={
"workspace_id": "550e8400-e29b-41d4-a716-446655440000",
"agent_type": "synthesis",
"limit": 10,
},
)
const resp = await fetch(
"https://api.praxiomai.xyz/api/admin/agent-runs?workspace_id=550e8400-...&agent_type=synthesis&limit=10",
{ headers: { Authorization: `Bearer ${token}` } }
);
Response (200 OK)
{
"runs": [
{
"id": "c3d4e5f6-...",
"workspace_id": "550e8400-...",
"agent_type": "synthesis",
"tool_name": "synthesize_research",
"input_data": {"source_ids": ["a1b2...", "b2c3..."]},
"output_data": {"insight_count": 5},
"latency_ms": 23450,
"tokens_used": 15000,
"cost_usd": 0.045,
"status": "success",
"error_message": null,
"executed_at": "2026-03-27T10:35:00Z"
}
],
"total": 1,
"limit": 10,
"offset": 0
}
Get Agent Run Stats
GET /api/admin/agent-runs/stats
Get aggregated agent run statistics for a workspace.
Query Parameters
| Name | Type | Required | Description |
|---|---|---|---|
workspace_id | UUID | Yes | Target workspace |
Example Request
curl "https://api.praxiomai.xyz/api/admin/agent-runs/stats?workspace_id=550e8400-..." \
-H "Authorization: Bearer $TOKEN"
Response (200 OK)
{
"total_runs": 47,
"total_cost_usd": 2.15,
"runs_by_type": {
"synthesis": 12,
"recommendation": 8,
"drafting": 5,
"chat": 22
},
"runs_by_status": {
"success": 44,
"failed": 2,
"partial": 1
}
}
Quality: RQS Metrics
GET /api/admin/quality/rqs
Get Research Quality Score metrics and trends.
Query Parameters
| Name | Type | Required | Description |
|---|---|---|---|
workspace_id | UUID | Yes | Target workspace |
days | integer | No | Time window in days (default: 30, options: 7, 30, 90) |
Response (200 OK)
{
"avg_rqs": 0.74,
"contract_pass_rate": 0.86,
"verifier_pass_rate": 0.81,
"daily_trend": [
{
"date": "2026-04-07",
"avg_rqs": 0.76,
"contract_pass_pct": 0.88,
"verifier_pass_pct": 0.82
}
],
"by_agent_type": [
{
"agent_type": "synthesis",
"avg_rqs": 0.79
}
],
"worst_workspaces": [
{
"workspace_name": "Acme Dashboard",
"avg_rqs": 0.42,
"run_count": 15,
"verifier_fail_count": 6
}
]
}
Quality: User Feedback
GET /api/admin/quality/feedback
Get user feedback sentiment analysis.
Query Parameters
| Name | Type | Required | Description |
|---|---|---|---|
workspace_id | UUID | Yes | Target workspace |
days | integer | No | Time window in days (default: 30) |
Response (200 OK)
{
"positive_rate": 0.72,
"avg_rating": 4.1,
"rated_count": 85,
"unrated_count": 120,
"daily_trend": [
{
"date": "2026-04-07",
"positive_count": 8,
"negative_count": 3
}
],
"top_positive_tags": [
{ "tag": "accurate", "count": 24 }
],
"top_negative_tags": [
{ "tag": "incomplete", "count": 7 }
],
"worst_workspaces": [
{
"workspace_name": "Acme Dashboard",
"positive_rate": 0.45,
"total_rated": 20
}
]
}
Quality: Entropy (Data Quality)
GET /api/admin/quality/entropy
Get entropy scanning results showing data quality issues across workspaces.
Query Parameters
| Name | Type | Required | Description |
|---|---|---|---|
workspace_id | UUID | Yes | Target workspace |
Response (200 OK)
{
"total_active_issues": 12,
"high_severity_count": 3,
"affected_workspaces": 2,
"by_type": [
{
"issue_type": "stale_source",
"severity": "high",
"count": 2
},
{
"issue_type": "contradictory_insights",
"severity": "medium",
"count": 5
}
],
"worst_workspaces": [
{
"workspace_name": "Acme Dashboard",
"high_count": 2,
"medium_count": 3,
"low_count": 1,
"last_scanned_at": "2026-04-07T02:15:00Z"
}
],
"recent_scans": [
{
"workspace_name": "Acme Dashboard",
"scanned_at": "2026-04-07T02:15:00Z",
"issue_count": 6,
"high_severity_count": 2
}
]
}
Admin Portal UI
The admin portal at admin.praxiomai.xyz provides three main views:
Conversations View
A filterable, paginated list of all conversations across workspaces. Clicking a conversation opens the ConversationSlideOver forensics panel showing:
- Full message history with tool call details
- Agent run metadata (model, cost, duration, tokens)
- Quality scores (RQS, contract pass/fail, verifier score)
- User feedback (thumbs up/down, tags)
Quality View
Three tabs for monitoring quality:
| Tab | Contents |
|---|---|
| RQS and Verifier | Average RQS, contract pass rate, verifier pass rate, daily trend table, breakdown by agent type, worst workspaces |
| User Feedback | Positive rate, average rating, daily trend, top positive/negative tags, worst workspaces |
| Entropy / Data Quality | Total active issues, severity breakdown, issues by type, worst workspaces, recent scan history |
All tabs support time window filtering (7, 30, or 90 days).
Was this helpful?