{% extends "base.html" %} {% block title %}Prompt Analytics - AISBF Dashboard{% endblock %} {% macro format_tokens(value) %} {% if value is none or value == 0 %}0{% else %} {% set val = value | float %} {% if val >= 1000000000 %} {{ "%.2f"|format(val / 1000000000) }}B {% elif val >= 1000000 %} {{ "%.2f"|format(val / 1000000) }}M {% elif val >= 1000 %} {{ "%.2f"|format(val / 1000) }}K {% else %} {{ "%.0f"|format(val) }} {% endif %} {% endif %} {% endmacro %} {% macro pct(part, total) %} {% if total and total > 0 %} {{ "%.0f"|format((part / total) * 100) }}% {% else %} 0% {% endif %} {% endmacro %} {% block content %}

Prompt Analytics

Native AISBF prompt analysis for prompt-security findings, request composition, blocking posture, and model/provider exposure. {% if is_config_admin %} This view can compare global traffic with user-scoped runs when filters are applied. {% else %} This view is limited to your own requests and does not expose global visitor activity. {% endif %}

📊 Analytics 🧠 Prompt Analytics {% if is_config_admin %} 👣 Traffic & Visits {% endif %} 💾 Response Cache ⏱️ Rate Limits

Filter by Date Range

Filter by Provider, Model, Rotation, Autoselect, or User

{% if from_date %}{% endif %} {% if to_date %}{% endif %}
{% if is_admin %}
{% endif %}
{% set details = prompt_analysis_details or {} %} {% set posture = details.posture or {} %} {% set total_runs = prompt_analysis_overview.total_runs if prompt_analysis_overview else 0 %} {% if prompt_analysis_overview and prompt_analysis_overview.total_runs > 0 %}
Analyzed Requests
{{ prompt_analysis_overview.total_runs }}
Blocked Requests
{{ prompt_analysis_overview.blocked_runs }}
{{ pct(prompt_analysis_overview.blocked_runs, prompt_analysis_overview.total_runs) }} of analyzed runs
High Risk
{{ prompt_analysis_overview.high_risk_runs }}
{{ pct(prompt_analysis_overview.high_risk_runs, prompt_analysis_overview.total_runs) }} flagged as high risk
Avg Prompt / Context Tokens
{{ "%.0f"|format(prompt_analysis_overview.avg_prompt_tokens) }} / {{ "%.0f"|format(prompt_analysis_overview.avg_effective_context) }}
Prompt body versus effective context footprint
Runs With Tools
{{ posture.runs_with_tools or 0 }}
{{ pct(posture.runs_with_tools or 0, total_runs) }} included tool definitions
Runs With System Prompt
{{ posture.runs_with_system_prompt or 0 }}
{{ pct(posture.runs_with_system_prompt or 0, total_runs) }} had an explicit system layer
Avg Findings / Avg Risk Score
{{ "%.1f"|format(posture.avg_findings_count or 0) }} / {{ "%.0f"|format(posture.avg_risk_score or 0) }}
Average detector density per analyzed request

Prompt Risk Timeline

Run volume, high-risk detections, and blocking decisions over time.

Scope

Visible Dataset
{% if is_config_admin %}{% if global_only == '1' %}Global-only{% elif selected_user %}Selected user{% else %}All visible scopes{% endif %}{% else %}Current user only{% endif %}
{% if is_config_admin %} Apply the user filter or enable Global requests only to separate user-owned prompt analysis from global admin traffic. {% else %} Prompt analytics are already scoped to your own requests, with no traffic or visitor telemetry exposed here. {% endif %}

Top Prompt Findings

{% if prompt_analysis_overview.findings %} {% for finding in prompt_analysis_overview.findings %} {% endfor %}
CategorySeverityCount
{{ finding.category }}{{ finding.severity }}{{ finding.count }}
{% else %}

No findings were recorded in this range.

{% endif %}

Risk Level Mix

{% if details.risk_breakdown %} {% for row in details.risk_breakdown %} {% endfor %}
RiskRunsShare
{{ row.risk_level }}{{ row.count }}{{ pct(row.count, total_runs) }}
{% else %}

No risk distribution available.

{% endif %}

Providers Under Prompt Pressure

{% if details.provider_breakdown %} {% for provider in details.provider_breakdown %} {% endfor %}
ProviderRunsBlockedHigh Risk
{{ provider.provider_id }} {{ provider.runs }} {{ provider.blocked }} {{ provider.high_risk }}
{% else %}

No provider breakdown available.

{% endif %}

Most Analyzed Models

{% if details.model_breakdown %} {% for model in details.model_breakdown %} {% endfor %}
ModelRunsBlockedHigh Risk
{{ model.model_name }} {{ model.runs }} {{ model.blocked }} {{ model.high_risk }}
{% else %}

No model breakdown available.

{% endif %}

Dominant Prompt Segment

{% if details.segment_breakdown %} {% for row in details.segment_breakdown %} {% endfor %}
SegmentRunsShare
{{ row.role }}{{ row.count }}{{ pct(row.count, total_runs) }}
{% else %}

No segment mix available.

{% endif %}

Prompt Shapes

{% if details.shape_breakdown %} {% for row in details.shape_breakdown %} {% endfor %}
ShapeRunsShare
{{ row.prompt_shape }}{{ row.count }}{{ pct(row.count, total_runs) }}
{% else %}

No prompt-shape data available.

{% endif %}

Recent Prompt Analysis Runs

{% if details.recent_runs %}
{% for run in details.recent_runs %} {% endfor %}
When Provider Model Risk Blocked Prompt / Context Shape Signals Findings
{{ run.created_at }} {{ run.provider_id }}
{{ run.model_name }}
{% if run.rotation_id or run.autoselect_id %}
{% if run.rotation_id %}Rotation: {{ run.rotation_id }}{% endif %} {% if run.rotation_id and run.autoselect_id %} | {% endif %} {% if run.autoselect_id %}Autoselect: {{ run.autoselect_id }}{% endif %}
{% endif %}
{{ run.risk_level }}
score {{ run.risk_score }}
 {% if run.blocked %}Yes{% else %}No{% endif %} {{ format_tokens(run.prompt_tokens) }} / {{ format_tokens(run.effective_context) }}
{{ run.prompt_shape }}
largest: {{ run.largest_segment_role }}
 {% if run.has_system_prompt %}system{% endif %} {% if run.has_tools %}tools{% endif %} {% if not run.has_system_prompt and not run.has_tools %}base{% endif %}
{{ run.findings_count }} total
H {{ run.high_count }} / M {{ run.medium_count }} / I {{ run.info_count }}
{% if run.findings %}
{% for finding in run.findings %}
{{ finding.category }} ({{ finding.severity }}) x{{ finding.count }}
{% endfor %}
{% endif %}
{% else %}

No recent prompt-analysis runs matched the current filters.

{% endif %}
{% else %}

No prompt analysis data available yet.

Prompt analytics are only recorded when Prompt Security or Context Lens Analytics is enabled for the active request path.
The shipped defaults keep both features disabled. Enable them from Dashboard → Settings or via provider/model overrides, then send new requests through AISBF.
Optional related controls include Block High-Risk Prompts, NSFW Classification, and Privacy Classification.
{% endif %}
Back to Dashboard Analytics {% if is_admin %} Rate Limits Response Cache {% endif %}
{% endblock %}