chore(release): bump version to 1.2.6

- Add VERSION build arg to Dockerfile
- Pass --build-arg VERSION in release workflow
- Remove VERSION env override from docker-compose files
- Version is now immutable inside the image, no runtime env var needed

.env.example

@@ -33,3 +33,15 @@ SIEM_WEBHOOK_URL=
 
 # Optional: enable rule-based alerting during ingestion
 ALERTS_ENABLED=false
+
+# Optional: LLM configuration for natural language querying (/api/ask)
+# Supports any OpenAI-compatible API (OpenAI, Azure OpenAI, Ollama, etc.)
+# For Azure OpenAI / MS Foundry, set BASE_URL to your deployment endpoint
+# (e.g. https://your-resource.openai.azure.com/openai/deployments/your-deployment)
+# and set API_VERSION to something like 2025-01-01-preview
+LLM_API_KEY=
+LLM_BASE_URL=https://api.openai.com/v1
+LLM_MODEL=gpt-4o-mini
+LLM_MAX_EVENTS=200
+LLM_TIMEOUT_SECONDS=30
+LLM_API_VERSION=

.gitea/workflows/release.yml

@@ -16,7 +16,7 @@ jobs:
         run: echo "${{ secrets.REGISTRY_TOKEN }}" | docker login git.cqre.net -u ${{ github.actor }} --password-stdin 2>&1 | grep -v "WARNING! Your credentials are stored unencrypted"
 
       - name: Build Docker image
-        run: docker build ./backend --tag git.cqre.net/cqrenet/aoc-backend:${{ gitea.ref_name }}
+        run: docker build ./backend --build-arg VERSION=${{ gitea.ref_name }} --tag git.cqre.net/cqrenet/aoc-backend:${{ gitea.ref_name }}
 
       - name: Push Docker image
         run: docker push git.cqre.net/cqrenet/aoc-backend:${{ gitea.ref_name }}

DEPLOY.md

@@ -0,0 +1,103 @@
+# Production Deployment Guide
+
+## Overview
+
+AOC runs as a set of Docker containers orchestrated by Docker Compose:
+
+- **nginx** — reverse proxy, TLS termination, static file serving
+- **backend** — FastAPI application (Gunicorn + Uvicorn workers)
+- **mongo** — MongoDB data store (not exposed externally)
+
+## Prerequisites
+
+- Docker Engine 24+ and Docker Compose plugin
+- A server with ports 80/443 reachable from your users
+- TLS certificates (place in `nginx/ssl/` or use Let's Encrypt)
+- A valid `.env` file at the repo root (see `.env.example`)
+
+## Quick start
+
+1. **Clone / pull the latest release**
+
+   ```bash
+   git checkout v1.1.0
+   ```
+
+2. **Copy and edit environment variables**
+
+   ```bash
+   cp .env.example .env
+   # Edit .env and fill in real credentials
+   ```
+
+3. **Set the release version**
+
+   ```bash
+   export AOC_VERSION=v1.1.0
+   ```
+
+4. **Deploy**
+
+   ```bash
+   docker compose -f docker-compose.prod.yml pull
+   docker compose -f docker-compose.prod.yml up -d
+   ```
+
+5. **Verify**
+
+   ```bash
+   curl http://localhost/health
+   curl http://localhost/api/events
+   ```
+
+## Updating to a new release
+
+```bash
+export AOC_VERSION=v1.2.0
+docker compose -f docker-compose.prod.yml pull
+docker compose -f docker-compose.prod.yml up -d
+```
+
+## Enabling HTTPS
+
+### Option A: Use your own certificates
+
+1. Place `cert.pem` and `key.pem` in `nginx/ssl/`
+2. Uncomment the HTTPS server block in `nginx/nginx.conf`
+3. Uncomment the HTTP → HTTPS redirect server block
+4. Reload nginx:
+
+   ```bash
+   docker compose -f docker-compose.prod.yml exec nginx nginx -s reload
+   ```
+
+### Option B: Let's Encrypt with Certbot
+
+Replace the `nginx` service in `docker-compose.prod.yml` with a Certbot-friendly setup (e.g., use the `nginx-proxy` + `acme-companion` stack) or mount the Certbot certificates into `nginx/ssl/`.
+
+## Security hardening
+
+- MongoDB is **not exposed** to the host — only the backend container can reach it.
+- The backend runs as a non-root (`aoc`) user inside the container.
+- nginx adds security headers (`X-Frame-Options`, `X-Content-Type-Options`, etc.).
+- Keep `.env` out of version control — it is listed in `.gitignore`.
+
+## Rollback
+
+```bash
+export AOC_VERSION=v1.0.3
+docker compose -f docker-compose.prod.yml pull
+docker compose -f docker-compose.prod.yml up -d
+```
+
+## Monitoring
+
+- Prometheus metrics: `http://your-host/metrics`
+- Health check: `http://your-host/health`
+- Container logs:
+
+  ```bash
+  docker compose -f docker-compose.prod.yml logs -f backend
+  docker compose -f docker-compose.prod.yml logs -f nginx
+  docker compose -f docker-compose.prod.yml logs -f mongo
+  ```

RELEASE_NOTES_v1.1.0.md

@@ -0,0 +1,56 @@
+# AOC v1.1.0 Release Notes
+
+**Release date:** 2026-04-20
+
+## What's new
+
+### Natural language query (`/api/ask`)
+Ask questions in plain English and get AI-generated answers backed by your audit logs.
+
+- **Regex-based parsing** extracts time ranges (`last 3 days`, `yesterday`, `today`) and entities (`device ABC123`, `user bob@example.com`) without calling an LLM — fast and deterministic.
+- **AI narrative summarisation** via any OpenAI-compatible API (OpenAI, Azure OpenAI, MS Foundry, Ollama). The LLM reads the matching events and writes a concise story for non-expert admins.
+- **Graceful fallback** when no LLM is configured — returns a structured bullet list instead of a narrative.
+- **Cited evidence** — every answer includes the raw events that back it up, so admins can verify claims.
+
+### Azure OpenAI / MS Foundry support
+- Automatic `api-key` header detection for Azure endpoints.
+- `LLM_API_VERSION` config for Azure `api-version` query parameters.
+- `max_completion_tokens` support for newer model deployments.
+
+### Production hardening
+- **Dockerfile:** runs as non-root user, uses Gunicorn + Uvicorn workers.
+- **docker-compose.prod.yml:** MongoDB is internal-only (no host port exposure), health checks on all services, nginx reverse proxy with security headers.
+- **nginx config:** gzip, security headers (`X-Frame-Options`, `X-Content-Type-Options`), ready for TLS.
+
+### Frontend
+- New **"Ask a question"** panel at the top of the page.
+- Markdown rendering for LLM answers (bold, italic, code).
+- Orange warning banner when LLM is not configured or fails.
+
+### Tests
+- 29 new tests covering ask parsing, query building, and endpoint behaviour.
+- 62 tests total, all passing.
+
+## Configuration
+
+Add to your `.env`:
+
+```bash
+# Required for AI narrative summarisation
+LLM_API_KEY=your-key
+LLM_BASE_URL=https://api.openai.com/v1
+LLM_MODEL=gpt-4o-mini
+LLM_MAX_EVENTS=50
+LLM_TIMEOUT_SECONDS=30
+LLM_API_VERSION=                 # set for Azure OpenAI, e.g. 2024-12-01-preview
+```
+
+## Upgrade notes
+
+No breaking changes. Existing `/api/events`, filters, pagination, tags, and comments work unchanged.
+
+## Docker image
+
+```
+git.cqre.net/cqrenet/aoc-backend:v1.1.0
+```

RELEASE_NOTES_v1.2.5.md

@@ -0,0 +1,78 @@
+# AOC v1.2.5 Release Notes
+
+**Release date:** 2026-04-20
+
+---
+
+## What's new
+
+### Natural language query (`/api/ask`)
+Ask questions in plain English and get AI-generated answers backed by your audit logs.
+
+- **Regex-based parsing** extracts time ranges (`last 3 days`, `yesterday`, `today`) and entities (`device ABC123`, `user bob@example.com`) without calling an LLM.
+- **AI narrative summarisation** via any OpenAI-compatible API (OpenAI, Azure OpenAI, MS Foundry, Ollama).
+- **Graceful fallback** when no LLM is configured — returns a structured bullet list with a clear error banner.
+- **Cited evidence** — every answer includes the raw events that back it up.
+
+### Filter-aware queries
+The ask endpoint now respects the filter panel. When you set **Service = Exchange**, **Result = failure** and ask *"What happened to device X?"*, the LLM only sees failed Exchange events for that device.
+
+### Scales to thousands of events
+For large result sets (>50 events), the LLM receives an **aggregated overview** instead of a raw dump:
+- Counts by service, action, result, and actor
+- Failure highlights
+- The 50 most recent raw events as samples
+
+This keeps token usage low while preserving accuracy.
+
+### Azure OpenAI / MS Foundry support
+- Automatic `api-key` header detection for Azure endpoints.
+- `LLM_API_VERSION` config for Azure `api-version` query parameters.
+- `max_completion_tokens` support for newer model deployments.
+
+### Version display
+- `GET /api/version` endpoint reads the `VERSION` file.
+- Frontend shows a version badge in the header (e.g., **1.2.5**).
+
+### Production hardening (from v1.1.0)
+- Dockerfile runs as non-root user with Gunicorn + Uvicorn workers.
+- `docker-compose.prod.yml` with internal-only MongoDB, health checks, and nginx reverse proxy.
+- Security headers (`X-Frame-Options`, `X-Content-Type-Options`, etc.).
+
+---
+
+## Configuration
+
+Add to your `.env`:
+
+```bash
+# Required for AI narrative summarisation
+LLM_API_KEY=your-key
+LLM_BASE_URL=https://api.openai.com/v1
+LLM_MODEL=gpt-4o-mini
+LLM_MAX_EVENTS=200
+LLM_TIMEOUT_SECONDS=30
+LLM_API_VERSION=                 # set for Azure OpenAI, e.g. 2024-12-01-preview
+```
+
+For Azure OpenAI / MS Foundry:
+```bash
+LLM_BASE_URL=https://your-resource.openai.azure.com/openai/deployments/your-deployment
+LLM_API_KEY=your-azure-key
+LLM_API_VERSION=2024-12-01-preview
+LLM_MODEL=your-deployment-name
+```
+
+---
+
+## Upgrade notes
+
+No breaking changes. Existing `/api/events`, filters, pagination, tags, and comments work unchanged.
+
+---
+
+## Docker image
+
+```
+git.cqre.net/cqrenet/aoc-backend:v1.2.5
+```

VERSION

@@ -1 +1 @@
-1.0.3
+1.2.6

backend/Dockerfile

@@ -1,6 +1,31 @@
 FROM python:3.11-slim
+
+# Bake the version into the image at build time
+ARG VERSION=unknown
+ENV VERSION=${VERSION}
+
+# Security: run as non-root
+RUN groupadd -r aoc && useradd -r -g aoc aoc
+
 WORKDIR /app
+
+# Install dependencies first for layer caching
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy application code
 COPY . .
-CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]
+
+# Create directories for potential volume mounts and fix permissions
+RUN mkdir -p /app/data && chown -R aoc:aoc /app
+
+USER aoc
+
+# Production: use gunicorn with uvicorn workers
+# Workers = 2-4 x $NUM_CORES; keep it conservative for containerised workloads
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONDONTWRITEBYTECODE=1
+
+EXPOSE 8000
+
+CMD ["gunicorn", "main:app", "-k", "uvicorn.workers.UvicornWorker", "--bind", "0.0.0.0:8000", "--workers", "2", "--timeout", "120", "--access-logfile", "-", "--error-logfile", "-"]

backend/config.py

@@ -42,6 +42,14 @@ class Settings(BaseSettings):
     # Alerting
     ALERTS_ENABLED: bool = False
 
+    # LLM / Natural Language Query
+    LLM_API_KEY: str = ""
+    LLM_BASE_URL: str = "https://api.openai.com/v1"
+    LLM_MODEL: str = "gpt-4o-mini"
+    LLM_MAX_EVENTS: int = 200
+    LLM_TIMEOUT_SECONDS: int = 30
+    LLM_API_VERSION: str = ""  # e.g. 2025-01-01-preview for Azure OpenAI
+
 
 _settings = Settings()
 
@@ -68,3 +76,10 @@ CORS_ORIGINS = [o.strip() for o in _settings.CORS_ORIGINS.split(",") if o.strip(
 SIEM_ENABLED = _settings.SIEM_ENABLED
 SIEM_WEBHOOK_URL = _settings.SIEM_WEBHOOK_URL
 ALERTS_ENABLED = _settings.ALERTS_ENABLED
+
+LLM_API_KEY = _settings.LLM_API_KEY
+LLM_BASE_URL = _settings.LLM_BASE_URL
+LLM_MODEL = _settings.LLM_MODEL
+LLM_MAX_EVENTS = _settings.LLM_MAX_EVENTS
+LLM_TIMEOUT_SECONDS = _settings.LLM_TIMEOUT_SECONDS
+LLM_API_VERSION = _settings.LLM_API_VERSION

backend/frontend/index.html

@@ -12,7 +12,7 @@
   <div class="page" x-data="aocApp()" x-init="initApp()">
     <header class="hero">
       <div>
-        <p class="eyebrow">Admin Operations Center</p>
+        <p class="eyebrow">Admin Operations Center <span class="version-badge" x-text="appVersion"></span></p>
         <h1>Directory Audit Explorer</h1>
         <p class="lede">Filter Microsoft Entra audit events by user, app, time, action, and action type.</p>
       </div>
@@ -38,6 +38,49 @@
       </div>
     </section>
 
+    <section class="panel">
+      <h3>Ask a question</h3>
+      <form class="ask-form" @submit.prevent="askQuestion()">
+        <div class="ask-row">
+          <input
+            type="text"
+            placeholder="What happened to device ABC123 in the last 3 days?"
+            x-model="askQuestionText"
+            class="ask-input"
+          />
+          <button type="submit" :disabled="askLoading" x-text="askLoading ? 'Thinking…' : 'Ask'">Ask</button>
+        </div>
+        <div x-show="hasActiveFilters()" class="ask-filter-hint">
+          <small>Respecting active filters: <span x-text="activeFilterSummary()"></span></small>
+        </div>
+      </form>
+      <template x-if="askAnswer">
+        <div class="ask-result">
+          <div x-show="askLlmError" class="ask-error" x-text="askLlmError"></div>
+          <div class="ask-answer" x-html="askAnswerHtml"></div>
+          <template x-if="askEvents.length">
+            <div class="ask-events">
+              <h4>Referenced events</h4>
+              <template x-for="(evt, idx) in askEvents" :key="evt.id || idx">
+                <article class="event event--compact">
+                  <div class="event__meta">
+                    <span class="pill" x-text="evt.display_category || evt.service || '—'"></span>
+                    <span class="pill" :class="['success','succeeded','ok','passed'].includes((evt.result || '').toLowerCase()) ? 'pill--ok' : 'pill--warn'" x-text="evt.result || '—'"></span>
+                  </div>
+                  <h3 x-text="evt.operation || '—'"></h3>
+                  <p class="event__detail" x-show="evt.display_summary"><strong>Summary:</strong> <span x-text="evt.display_summary"></span></p>
+                  <p class="event__detail"><strong>Actor:</strong> <span x-text="evt.actor_display || '—'"></span></p>
+                  <p class="event__detail"><strong>Target:</strong> <span x-text="Array.isArray(evt.target_displays) ? evt.target_displays.join(', ') : '—'"></span></p>
+                  <p class="event__detail"><strong>When:</strong> <span x-text="evt.timestamp ? new Date(evt.timestamp).toLocaleString() : '—'"></span></p>
+                </article>
+              </template>
+            </div>
+          </template>
+          <button type="button" class="ghost" @click="clearAsk()">Clear</button>
+        </div>
+      </template>
+    </section>
+
     <section class="panel">
       <form id="filters" class="filters" @submit.prevent="resetPagination(); loadEvents()">
         <div class="filter-row">
@@ -200,8 +243,17 @@
           actor: '', selectedServices: [], search: '', operation: '', result: '', start: '', end: '', limit: 100, includeTags: '', excludeTags: '',
         },
         options: { actors: [], services: [], operations: [], results: [] },
+        appVersion: '',
+        askQuestionText: '',
+        askLoading: false,
+        askAnswer: '',
+        askAnswerHtml: '',
+        askEvents: [],
+        askLlmUsed: false,
+        askLlmError: '',
 
         async initApp() {
+          await this.loadVersion();
           await this.initAuth();
           if (!this.authConfig?.auth_enabled || this.accessToken) {
             await this.loadFilterOptions();
@@ -210,6 +262,16 @@
           }
         },
 
+        async loadVersion() {
+          try {
+            const res = await fetch('/api/version');
+            if (res.ok) {
+              const body = await res.json();
+              this.appVersion = body.version || '';
+            }
+          } catch {}
+        },
+
         authHeader() {
           return this.accessToken ? { Authorization: `Bearer ${this.accessToken}` } : {};
         },
@@ -437,6 +499,93 @@
           this.loadEvents();
         },
 
+        async askQuestion() {
+          const q = this.askQuestionText.trim();
+          if (!q) return;
+          this.askLoading = true;
+          this.askAnswer = '';
+          this.askAnswerHtml = '';
+          this.askEvents = [];
+          this.askLlmError = '';
+
+          const payload = { question: q };
+          if (this.filters.selectedServices && this.filters.selectedServices.length) {
+            payload.services = this.filters.selectedServices;
+          }
+          if (this.filters.actor) payload.actor = this.filters.actor;
+          if (this.filters.operation) payload.operation = this.filters.operation;
+          if (this.filters.result) payload.result = this.filters.result;
+          if (this.filters.start) payload.start = new Date(this.filters.start).toISOString();
+          if (this.filters.end) payload.end = new Date(this.filters.end).toISOString();
+          if (this.filters.includeTags) {
+            payload.include_tags = this.filters.includeTags.split(/[,;]+/).map(t => t.trim()).filter(Boolean);
+          }
+          if (this.filters.excludeTags) {
+            payload.exclude_tags = this.filters.excludeTags.split(/[,;]+/).map(t => t.trim()).filter(Boolean);
+          }
+
+          try {
+            const res = await fetch('/api/ask', {
+              method: 'POST',
+              headers: { 'Content-Type': 'application/json', ...this.authHeader() },
+              body: JSON.stringify(payload),
+            });
+            if (!res.ok) throw new Error(await res.text());
+            const body = await res.json();
+            this.askAnswer = body.answer;
+            this.askAnswerHtml = this._mdToHtml(body.answer);
+            this.askEvents = body.events || [];
+            this.askLlmUsed = body.llm_used;
+            this.askLlmError = body.llm_error || '';
+          } catch (err) {
+            this.askAnswer = 'Sorry, something went wrong: ' + (err.message || 'Unknown error');
+            this.askAnswerHtml = this.askAnswer;
+          } finally {
+            this.askLoading = false;
+          }
+        },
+
+        clearAsk() {
+          this.askQuestionText = '';
+          this.askAnswer = '';
+          this.askAnswerHtml = '';
+          this.askEvents = [];
+          this.askLlmUsed = false;
+          this.askLlmError = '';
+        },
+
+        _mdToHtml(text) {
+          // Very lightweight markdown-to-HTML for LLM answers
+          return text
+            .replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;')
+            .replace(/\*\*(.+?)\*\*/g, '<strong>$1</strong>')
+            .replace(/\*(.+?)\*/g, '<em>$1</em>')
+            .replace(/`([^`]+)`/g, '<code>$1</code>')
+            .replace(/Event #(\d+)/g, '<strong>Event #$1</strong>')
+            .replace(/\n/g, '<br>');
+        },
+
+        hasActiveFilters() {
+          return this.filters.actor || this.filters.operation || this.filters.result ||
+            this.filters.start || this.filters.end || this.filters.includeTags ||
+            this.filters.excludeTags ||
+            (this.filters.selectedServices && this.filters.selectedServices.length &&
+             this.filters.selectedServices.length < this.options.services.length);
+        },
+
+        activeFilterSummary() {
+          const parts = [];
+          if (this.filters.actor) parts.push('actor');
+          if (this.filters.operation) parts.push('action');
+          if (this.filters.result) parts.push('result');
+          if (this.filters.start || this.filters.end) parts.push('time');
+          if (this.filters.includeTags || this.filters.excludeTags) parts.push('tags');
+          const svcCount = this.filters.selectedServices?.length || 0;
+          const allCount = this.options.services?.length || 0;
+          if (svcCount && svcCount < allCount) parts.push(`${svcCount} service${svcCount === 1 ? '' : 's'}`);
+          return parts.join(', ') || 'none';
+        },
+
         async bulkTagMatching() {
           const tag = prompt('Enter tag to apply to all matching events:');
           if (!tag || !tag.trim()) return;

backend/frontend/style.css

@@ -377,6 +377,113 @@ input {
   margin: 0;
 }
 
+/* Ask / Natural Language Query */
+.ask-form {
+  margin-top: 10px;
+}
+
+.ask-row {
+  display: flex;
+  gap: 10px;
+  align-items: center;
+}
+
+.ask-input {
+  flex: 1;
+  padding: 12px 14px;
+  border-radius: 10px;
+  border: 1px solid var(--border);
+  background: rgba(255, 255, 255, 0.02);
+  color: var(--text);
+  font-size: 15px;
+}
+
+.ask-result {
+  margin-top: 16px;
+}
+
+.ask-answer {
+  background: rgba(125, 211, 252, 0.06);
+  border: 1px solid rgba(125, 211, 252, 0.2);
+  border-radius: 12px;
+  padding: 16px;
+  line-height: 1.55;
+  margin-bottom: 14px;
+}
+
+.ask-answer code {
+  background: rgba(255,255,255,0.06);
+  padding: 2px 6px;
+  border-radius: 6px;
+  font-size: 13px;
+}
+
+.ask-error {
+  background: rgba(249, 115, 22, 0.1);
+  border: 1px solid rgba(249, 115, 22, 0.3);
+  border-radius: 8px;
+  padding: 10px 14px;
+  color: #fdba74;
+  font-size: 14px;
+  margin-bottom: 10px;
+}
+
+.ask-filter-hint {
+  margin-top: 6px;
+  color: var(--muted);
+}
+
+.version-badge {
+  display: inline-block;
+  margin-left: 8px;
+  padding: 2px 8px;
+  border-radius: 999px;
+  background: rgba(125, 211, 252, 0.15);
+  border: 1px solid rgba(125, 211, 252, 0.3);
+  color: var(--accent-strong);
+  font-size: 11px;
+  font-weight: 600;
+  letter-spacing: 0.05em;
+  vertical-align: middle;
+}
+
+.ask-events {
+  margin-bottom: 14px;
+}
+
+.ask-events h4 {
+  margin: 0 0 10px;
+  color: var(--muted);
+  font-size: 14px;
+  text-transform: uppercase;
+  letter-spacing: 0.06em;
+}
+
+.event--compact {
+  padding: 12px;
+  margin-bottom: 10px;
+}
+
+.event--compact h3 {
+  font-size: 15px;
+}
+
+.source-health {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(min(200px, 100%), 1fr));
+  gap: 10px;
+}
+
+.health-card {
+  border: 1px solid var(--border);
+  border-radius: 10px;
+  padding: 10px 12px;
+  background: rgba(255, 255, 255, 0.02);
+  display: flex;
+  flex-direction: column;
+  gap: 4px;
+}
+
 @media (max-width: 640px) {
   .hero {
     flex-direction: column;
@@ -386,4 +493,9 @@ input {
     flex-direction: column;
     align-items: stretch;
   }
+
+  .ask-row {
+    flex-direction: column;
+    align-items: stretch;
+  }
 }

backend/main.py

@@ -14,6 +14,7 @@ from fastapi.responses import Response
 from fastapi.staticfiles import StaticFiles
 from metrics import observe_request, prometheus_metrics
 from middleware import CorrelationIdMiddleware
+from routes.ask import router as ask_router
 from routes.config import router as config_router
 from routes.events import router as events_router
 from routes.fetch import router as fetch_router
@@ -112,6 +113,7 @@ app.include_router(events_router, prefix="/api")
 app.include_router(config_router, prefix="/api")
 app.include_router(webhooks_router, prefix="/api")
 app.include_router(health_router, prefix="/api")
+app.include_router(ask_router, prefix="/api")
 app.include_router(rules_router, prefix="/api")
 
 
@@ -132,6 +134,13 @@ async def metrics():
     return Response(content=prometheus_metrics(), media_type="text/plain")
 
 
+@app.get("/api/version")
+async def version():
+    import os
+
+    return {"version": os.environ.get("VERSION", "unknown")}
+
+
 frontend_dir = Path(__file__).parent / "frontend"
 app.mount("/", StaticFiles(directory=frontend_dir, html=True), name="frontend")
 

backend/models/api.py

@@ -70,3 +70,34 @@ class AlertRuleResponse(BaseModel):
     severity: str
     conditions: list[dict]
     message: str
+
+
+class AskRequest(BaseModel):
+    question: str
+    services: list[str] | None = None
+    actor: str | None = None
+    operation: str | None = None
+    result: str | None = None
+    start: str | None = None
+    end: str | None = None
+    include_tags: list[str] | None = None
+    exclude_tags: list[str] | None = None
+
+
+class AskEventRef(BaseModel):
+    id: str | None = None
+    timestamp: str | None = None
+    operation: str | None = None
+    actor_display: str | None = None
+    target_displays: list[str] | None = None
+    display_summary: str | None = None
+    service: str | None = None
+    result: str | None = None
+
+
+class AskResponse(BaseModel):
+    answer: str
+    events: list[AskEventRef]
+    query_info: dict
+    llm_used: bool
+    llm_error: str | None = None

backend/requirements.txt

@@ -11,3 +11,5 @@ pydantic-settings
 structlog
 tenacity
 prometheus-client
+httpx
+gunicorn

backend/routes/ask.py

@@ -0,0 +1,422 @@
+import json
+import re
+from datetime import UTC, datetime, timedelta
+
+import httpx
+import structlog
+from auth import require_auth
+from config import LLM_API_KEY, LLM_API_VERSION, LLM_BASE_URL, LLM_MAX_EVENTS, LLM_MODEL, LLM_TIMEOUT_SECONDS
+from database import events_collection
+from fastapi import APIRouter, Depends, HTTPException
+from models.api import AskRequest, AskResponse
+
+router = APIRouter(dependencies=[Depends(require_auth)])
+logger = structlog.get_logger("aoc.ask")
+
+# ---------------------------------------------------------------------------
+# Time-range extraction
+# ---------------------------------------------------------------------------
+
+_TIME_PATTERNS = [
+    (r"\blast\s+(\d+)\s+days?\b", "days"),
+    (r"\blast\s+(\d+)\s+hours?\b", "hours"),
+    (r"\blast\s+(\d+)\s+minutes?\b", "minutes"),
+    (r"\blast\s+week\b", "week"),
+    (r"\byesterday\b", "yesterday"),
+    (r"\btoday\b", "today"),
+    (r"\bin\s+the\s+last\s+(\d+)\s+days?\b", "days"),
+    (r"\bin\s+the\s+last\s+(\d+)\s+hours?\b", "hours"),
+]
+
+
+def _extract_time_range(question: str) -> tuple[str | None, str | None]:
+    """Return (start_iso, end_iso) or (None, None) if no time detected."""
+    now = datetime.now(UTC)
+    q_lower = question.lower()
+
+    for pattern, unit in _TIME_PATTERNS:
+        m = re.search(pattern, q_lower)
+        if not m:
+            continue
+        if unit == "week":
+            start = now - timedelta(days=7)
+        elif unit == "yesterday":
+            start = now - timedelta(days=1)
+        elif unit == "today":
+            start = now.replace(hour=0, minute=0, second=0, microsecond=0)
+        else:
+            num = int(m.group(1))
+            delta = {"days": timedelta(days=num), "hours": timedelta(hours=num), "minutes": timedelta(minutes=num)}[
+                unit
+            ]
+            start = now - delta
+        return start.isoformat().replace("+00:00", "Z"), now.isoformat().replace("+00:00", "Z")
+
+    return None, None
+
+
+# ---------------------------------------------------------------------------
+# Entity extraction
+# ---------------------------------------------------------------------------
+
+_ENTITY_HINTS = [
+    r"device\s+['\"]?([^'\"\s]+)['\"]?",
+    r"user\s+['\"]?([^'\"\s]+)['\"]?",
+    r"laptop\s+['\"]?([^'\"\s]+)['\"]?",
+    r"vm\s+['\"]?([^'\"\s]+)['\"]?",
+    r"server\s+['\"]?([^'\"\s]+)['\"]?",
+    r"computer\s+['\"]?([^'\"\s]+)['\"]?",
+]
+
+_EMAIL_RE = re.compile(r"[\w.+-]+@[\w-]+\.[\w.-]+")
+
+
+def _extract_entity(question: str) -> str | None:
+    """Best-effort extraction of the device / user / entity name."""
+    q_lower = question.lower()
+
+    # Look for explicit hints: "device ABC123"
+    for pattern in _ENTITY_HINTS:
+        m = re.search(pattern, q_lower)
+        if m:
+            # Extract from the original question to preserve case
+            start, end = m.span(1)
+            return question[start:end].strip().rstrip("?.!,;:")
+
+    # Look for quoted strings
+    m = re.search(r'"([^"]{2,50})"', question)
+    if m:
+        return m.group(1).strip()
+    m = re.search(r"'([^']{2,50})'", question)
+    if m:
+        return m.group(1).strip()
+
+    # Look for email addresses
+    m = _EMAIL_RE.search(question)
+    if m:
+        return m.group(0)
+
+    return None
+
+
+# ---------------------------------------------------------------------------
+# MongoDB query builder
+# ---------------------------------------------------------------------------
+
+
+def _build_event_query(
+    entity: str | None,
+    start: str | None,
+    end: str | None,
+    services: list[str] | None = None,
+    actor: str | None = None,
+    operation: str | None = None,
+    result: str | None = None,
+    include_tags: list[str] | None = None,
+    exclude_tags: list[str] | None = None,
+) -> dict:
+    filters = []
+
+    if start or end:
+        time_filter = {}
+        if start:
+            time_filter["$gte"] = start
+        if end:
+            time_filter["$lte"] = end
+        filters.append({"timestamp": time_filter})
+
+    if entity:
+        entity_safe = re.escape(entity)
+        filters.append(
+            {
+                "$or": [
+                    {"target_displays": {"$elemMatch": {"$regex": entity_safe, "$options": "i"}}},
+                    {"actor_display": {"$regex": entity_safe, "$options": "i"}},
+                    {"actor_upn": {"$regex": entity_safe, "$options": "i"}},
+                    {"raw_text": {"$regex": entity_safe, "$options": "i"}},
+                ]
+            }
+        )
+
+    if services:
+        filters.append({"service": {"$in": services}})
+    if actor:
+        actor_safe = re.escape(actor)
+        filters.append(
+            {
+                "$or": [
+                    {"actor_display": {"$regex": actor_safe, "$options": "i"}},
+                    {"actor_upn": {"$regex": actor_safe, "$options": "i"}},
+                    {"actor.user.userPrincipalName": {"$regex": actor_safe, "$options": "i"}},
+                ]
+            }
+        )
+    if operation:
+        filters.append({"operation": {"$regex": re.escape(operation), "$options": "i"}})
+    if result:
+        filters.append({"result": {"$regex": re.escape(result), "$options": "i"}})
+    if include_tags:
+        filters.append({"tags": {"$all": include_tags}})
+    if exclude_tags:
+        filters.append({"tags": {"$not": {"$all": exclude_tags}}})
+
+    return {"$and": filters} if filters else {}
+
+
+# ---------------------------------------------------------------------------
+# LLM summarisation
+# ---------------------------------------------------------------------------
+
+_SYSTEM_PROMPT = """You are an IT operations assistant. An administrator has asked a question about audit logs.
+Your job is to read the data below and write a concise, plain-language answer.
+
+The input may be either:
+- A small list of individual audit events (numbered Event #1, #2, etc.), or
+- An aggregated overview with counts by service, action, result, and actor, plus sample events.
+
+Rules:
+- Assume the reader is a non-expert admin.
+- For aggregated overviews: summarise the scale, top patterns, and highlight anomalies or failures.
+- For small event lists: group related events together and tell a coherent story.
+- Highlight anything unusual, failed actions, or privilege escalations.
+- Reference specific event numbers (e.g., "Event #3") when making claims so the user can verify.
+- If the data is an aggregated subset of a larger result set, acknowledge the scale (e.g., "847 events occurred — the top pattern was...").
+- If there are no events, say so clearly.
+- Keep the answer under 300 words.
+- Do not invent events or patterns that are not supported by the data.
+"""
+
+
+def _aggregate_counts(events: list[dict]) -> dict:
+    """Build lightweight aggregation tables for large result sets."""
+    from collections import Counter
+
+    svc_counts = Counter(e.get("service") or "Unknown" for e in events)
+    op_counts = Counter(e.get("operation") or "Unknown" for e in events)
+    result_counts = Counter(e.get("result") or "Unknown" for e in events)
+    actor_counts = Counter(e.get("actor_display") or "Unknown" for e in events)
+    return {
+        "services": svc_counts.most_common(10),
+        "operations": op_counts.most_common(10),
+        "results": result_counts.most_common(5),
+        "actors": actor_counts.most_common(10),
+    }
+
+
+def _format_events_for_llm(events: list[dict], total: int | None = None) -> str:
+    lines = []
+
+    # If we have a large result set, send aggregation + samples instead of raw dump
+    if total is not None and total > len(events) and len(events) >= 50:
+        lines.append(f"Result set overview: {total} total events (showing the {len(events)} most recent).\n")
+        agg = _aggregate_counts(events)
+        lines.append("Breakdown by service:")
+        for svc, cnt in agg["services"]:
+            lines.append(f"  {svc}: {cnt}")
+        lines.append("\nBreakdown by action:")
+        for op, cnt in agg["operations"]:
+            lines.append(f"  {op}: {cnt}")
+        lines.append("\nBreakdown by result:")
+        for res, cnt in agg["results"]:
+            lines.append(f"  {res}: {cnt}")
+        lines.append("\nTop actors:")
+        for actor, cnt in agg["actors"]:
+            lines.append(f"  {actor}: {cnt}")
+        # Include failures and a few recent samples
+        failures = [e for e in events if str(e.get("result") or "").lower() in ("failure", "failed")]
+        if failures:
+            lines.append(f"\nFailures ({len(failures)}):")
+            for e in failures[:10]:
+                ts = e.get("timestamp", "?")[:16].replace("T", " ")
+                op = e.get("operation", "unknown")
+                actor = e.get("actor_display", "unknown")
+                lines.append(f"  {ts} — {op} by {actor}")
+        lines.append("\nMost recent sample events:")
+    else:
+        if total is not None and total > len(events):
+            lines.append(f"Showing {len(events)} of {total} total matching events (most recent first):\n")
+
+    # Always include the first N raw events as detail (up to 50)
+    for i, e in enumerate(events[:50], 1):
+        ts = e.get("timestamp") or "unknown time"
+        op = e.get("operation") or "unknown action"
+        actor = e.get("actor_display") or "unknown actor"
+        targets = ", ".join(e.get("target_displays") or []) or "unknown target"
+        svc = e.get("service") or "unknown service"
+        result = e.get("result") or "unknown result"
+        summary = e.get("display_summary") or ""
+        lines.append(
+            f"Event #{i}\n"
+            f"  Time: {ts}\n"
+            f"  Service: {svc}\n"
+            f"  Action: {op}\n"
+            f"  Actor: {actor}\n"
+            f"  Target: {targets}\n"
+            f"  Result: {result}\n"
+            f"  Summary: {summary}\n"
+        )
+    return "\n".join(lines)
+
+
+def _build_chat_url(base_url: str, api_version: str) -> str:
+    """Construct the chat completions URL, handling Azure OpenAI endpoints."""
+    base = base_url.rstrip("/")
+    url = base if base.endswith("/chat/completions") else f"{base}/chat/completions"
+    if api_version:
+        url = f"{url}?api-version={api_version}"
+    return url
+
+
+async def _call_llm(question: str, events: list[dict], total: int | None = None) -> str:
+    if not LLM_API_KEY:
+        raise RuntimeError("LLM_API_KEY not configured")
+
+    context = _format_events_for_llm(events, total=total)
+    messages = [
+        {"role": "system", "content": _SYSTEM_PROMPT},
+        {
+            "role": "user",
+            "content": f"Question: {question}\n\nAudit events:\n{context}\n\nPlease answer the question based only on the events above.",
+        },
+    ]
+
+    url = _build_chat_url(LLM_BASE_URL, LLM_API_VERSION)
+    headers = {
+        "Content-Type": "application/json",
+    }
+    # Azure OpenAI uses api-key header; standard OpenAI uses Bearer token
+    if "azure" in LLM_BASE_URL.lower() or "cognitiveservices" in LLM_BASE_URL.lower():
+        headers["api-key"] = LLM_API_KEY
+    else:
+        headers["Authorization"] = f"Bearer {LLM_API_KEY}"
+
+    payload = {
+        "model": LLM_MODEL,
+        "messages": messages,
+        "max_completion_tokens": 800,
+    }
+
+    async with httpx.AsyncClient(timeout=LLM_TIMEOUT_SECONDS) as client:
+        resp = await client.post(url, headers=headers, json=payload)
+        if resp.status_code >= 400:
+            body = resp.text
+            logger.error("LLM API error", status_code=resp.status_code, url=url, response_body=body)
+            raise RuntimeError(f"LLM API error {resp.status_code}: {body[:500]}")
+        data = resp.json()
+        return data["choices"][0]["message"]["content"].strip()
+
+
+# ---------------------------------------------------------------------------
+# API endpoint
+# ---------------------------------------------------------------------------
+
+
+def _to_event_ref(e: dict) -> dict:
+    return {
+        "id": e.get("id"),
+        "timestamp": e.get("timestamp"),
+        "operation": e.get("operation"),
+        "actor_display": e.get("actor_display"),
+        "target_displays": e.get("target_displays"),
+        "display_summary": e.get("display_summary"),
+        "service": e.get("service"),
+        "result": e.get("result"),
+    }
+
+
+@router.post("/ask", response_model=AskResponse)
+async def ask_question(body: AskRequest, user: dict = Depends(require_auth)):
+    question = body.question.strip()
+    if not question:
+        raise HTTPException(status_code=400, detail="Question is required")
+
+    start, end = _extract_time_range(question)
+    entity = _extract_entity(question)
+
+    # Default to last 7 days if no time range detected
+    if not start:
+        now = datetime.now(UTC)
+        start = (now - timedelta(days=7)).isoformat().replace("+00:00", "Z")
+        end = now.isoformat().replace("+00:00", "Z")
+
+    query = _build_event_query(
+        entity,
+        start,
+        end,
+        services=body.services,
+        actor=body.actor,
+        operation=body.operation,
+        result=body.result,
+        include_tags=body.include_tags,
+        exclude_tags=body.exclude_tags,
+    )
+
+    try:
+        total = events_collection.count_documents(query)
+        cursor = events_collection.find(query).sort([("timestamp", -1)]).limit(LLM_MAX_EVENTS)
+        events = list(cursor)
+    except Exception as exc:
+        logger.error("Failed to query events for ask", error=str(exc))
+        raise HTTPException(status_code=500, detail=f"Database query failed: {exc}") from exc
+
+    for e in events:
+        e["_id"] = str(e.get("_id", ""))
+
+    # If no events, return early
+    if not events:
+        return AskResponse(
+            answer="I couldn't find any audit events matching your question. Try broadening the time range or checking the spelling of the device/user name.",
+            events=[],
+            query_info={"entity": entity, "start": start, "end": end, "event_count": 0},
+            llm_used=False,
+            llm_error="LLM not used — no events found." if not LLM_API_KEY else None,
+        )
+
+    # Try LLM summarisation
+    answer = ""
+    llm_used = False
+    llm_error = None
+    if not LLM_API_KEY:
+        llm_error = "LLM_API_KEY is not configured. Set it in your .env to enable AI narrative summarisation."
+    else:
+        try:
+            answer = await _call_llm(question, events, total=total)
+            llm_used = True
+        except Exception as exc:
+            llm_error = f"LLM call failed: {exc}"
+            logger.warning("LLM call failed, falling back to structured summary", error=str(exc))
+
+    # Fallback: structured summary if LLM unavailable or failed
+    if not answer:
+        parts = [f"Found {len(events)} event(s)"]
+        if entity:
+            parts.append(f"related to **{entity}**")
+        parts.append(f"between {start[:10]} and {end[:10]}.\n")
+
+        for i, e in enumerate(events[:10], 1):
+            ts = e.get("timestamp", "?")[:16].replace("T", " ")
+            op = e.get("operation", "unknown action")
+            actor = e.get("actor_display", "unknown")
+            targets = ", ".join(e.get("target_displays") or []) or "—"
+            result = e.get("result", "—")
+            parts.append(f"{i}. **{ts}** — {op} by {actor} on {targets} ({result})")
+
+        if len(events) > 10:
+            parts.append(f"\n...and {len(events) - 10} more events.")
+
+        answer = "\n".join(parts)
+
+    return AskResponse(
+        answer=answer,
+        events=[_to_event_ref(e) for e in events],
+        query_info={
+            "entity": entity,
+            "start": start,
+            "end": end,
+            "event_count": len(events),
+            "total_matched": total,
+            "mongo_query": json.dumps(query, default=str),
+        },
+        llm_used=llm_used,
+        llm_error=llm_error,
+    )

backend/tests/conftest.py

@@ -24,6 +24,7 @@ def client(mock_events_collection, mock_watermarks_collection, monkeypatch):
     monkeypatch.setattr("database.events_collection", mock_events_collection)
     monkeypatch.setattr("routes.fetch.events_collection", mock_events_collection)
     monkeypatch.setattr("routes.events.events_collection", mock_events_collection)
+    monkeypatch.setattr("routes.ask.events_collection", mock_events_collection)
     monkeypatch.setattr("watermark.watermarks_collection", mock_watermarks_collection)
     monkeypatch.setattr("routes.health.watermarks_collection", mock_watermarks_collection)
     monkeypatch.setattr("routes.fetch.get_watermark", lambda source: None)

backend/tests/test_ask.py

@@ -0,0 +1,352 @@
+from datetime import UTC, datetime, timedelta
+
+from routes.ask import _build_event_query, _extract_entity, _extract_time_range
+
+# ---------------------------------------------------------------------------
+# Unit tests: time-range extraction
+# ---------------------------------------------------------------------------
+
+
+class TestExtractTimeRange:
+    def test_last_n_days(self):
+        start, end = _extract_time_range("What happened in the last 3 days?")
+        assert start is not None
+        assert end is not None
+        # Start should be roughly 3 days before end
+        start_dt = datetime.fromisoformat(start.replace("Z", "+00:00"))
+        end_dt = datetime.fromisoformat(end.replace("Z", "+00:00"))
+        delta = end_dt - start_dt
+        assert delta.days == 3
+
+    def test_last_n_hours(self):
+        start, end = _extract_time_range("Show me events in the last 24 hours")
+        start_dt = datetime.fromisoformat(start.replace("Z", "+00:00"))
+        end_dt = datetime.fromisoformat(end.replace("Z", "+00:00"))
+        delta = end_dt - start_dt
+        assert delta.total_seconds() == 24 * 3600
+
+    def test_last_week(self):
+        start, end = _extract_time_range("What happened last week?")
+        start_dt = datetime.fromisoformat(start.replace("Z", "+00:00"))
+        end_dt = datetime.fromisoformat(end.replace("Z", "+00:00"))
+        assert (end_dt - start_dt).days == 7
+
+    def test_yesterday(self):
+        start, end = _extract_time_range("Show me yesterday's events")
+        start_dt = datetime.fromisoformat(start.replace("Z", "+00:00"))
+        end_dt = datetime.fromisoformat(end.replace("Z", "+00:00"))
+        assert (end_dt - start_dt).days == 1
+
+    def test_today(self):
+        start, end = _extract_time_range("What happened today?")
+        start_dt = datetime.fromisoformat(start.replace("Z", "+00:00"))
+        # end_dt is not needed for this assertion
+        # Should be from midnight today to now
+        assert start_dt.hour == 0
+        assert start_dt.minute == 0
+        assert start_dt.second == 0
+
+    def test_no_time_pattern_returns_none(self):
+        start, end = _extract_time_range("What happened to device ABC?")
+        assert start is None
+        assert end is None
+
+    def test_last_n_minutes(self):
+        start, end = _extract_time_range("Show me events in the last 15 minutes")
+        start_dt = datetime.fromisoformat(start.replace("Z", "+00:00"))
+        end_dt = datetime.fromisoformat(end.replace("Z", "+00:00"))
+        assert (end_dt - start_dt).total_seconds() == 15 * 60
+
+
+# ---------------------------------------------------------------------------
+# Unit tests: entity extraction
+# ---------------------------------------------------------------------------
+
+
+class TestExtractEntity:
+    def test_device_hint(self):
+        assert _extract_entity("What happened to device LAPTOP-001?") == "LAPTOP-001"
+
+    def test_user_hint(self):
+        assert _extract_entity("Show me user alice@example.com") == "alice@example.com"
+
+    def test_laptop_hint(self):
+        assert _extract_entity("What did laptop HR-Desk-04 do?") == "HR-Desk-04"
+
+    def test_server_hint(self):
+        assert _extract_entity("Check server WEB-01") == "WEB-01"
+
+    def test_quoted_string(self):
+        assert _extract_entity('What happened to "Surface-Pro-7"?') == "Surface-Pro-7"
+
+    def test_single_quoted_string(self):
+        assert _extract_entity("What happened to 'VM-WEB-01' today?") == "VM-WEB-01"
+
+    def test_email_address(self):
+        assert _extract_entity("What did tomas.svensson@contoso.com do?") == "tomas.svensson@contoso.com"
+
+    def test_no_entity_returns_none(self):
+        assert _extract_entity("What happened in the last 3 days?") is None
+
+    def test_vm_hint(self):
+        assert _extract_entity("Show me vm APP-SERVER-02") == "APP-SERVER-02"
+
+    def test_computer_hint(self):
+        assert _extract_entity("What happened to computer DESK-123?") == "DESK-123"
+
+
+# ---------------------------------------------------------------------------
+# Unit tests: query builder
+# ---------------------------------------------------------------------------
+
+
+class TestBuildEventQuery:
+    def test_entity_only(self):
+        q = _build_event_query("ABC123", None, None)
+        assert "$and" in q
+        or_clause = q["$and"][0]["$or"]
+        assert any("target_displays" in c for c in or_clause)
+        assert any("actor_display" in c for c in or_clause)
+        assert any("raw_text" in c for c in or_clause)
+
+    def test_time_only(self):
+        q = _build_event_query(None, "2024-01-01T00:00:00Z", "2024-01-02T00:00:00Z")
+        assert q["$and"][0]["timestamp"]["$gte"] == "2024-01-01T00:00:00Z"
+        assert q["$and"][0]["timestamp"]["$lte"] == "2024-01-02T00:00:00Z"
+
+    def test_entity_and_time(self):
+        q = _build_event_query("DEV-01", "2024-01-01T00:00:00Z", "2024-01-02T00:00:00Z")
+        assert len(q["$and"]) == 2
+        assert "timestamp" in q["$and"][0] or "timestamp" in q["$and"][1]
+
+    def test_empty_returns_empty(self):
+        q = _build_event_query(None, None, None)
+        assert q == {}
+
+    def test_entity_is_escaped_for_regex(self):
+        q = _build_event_query("DEV.01", None, None)
+        # The dot should be escaped in the regex
+        or_clause = q["$and"][0]["$or"]
+        raw_regex = or_clause[-1]["raw_text"]["$regex"]
+        assert raw_regex == "DEV\\.01"
+
+
+# ---------------------------------------------------------------------------
+# Integration tests: /api/ask endpoint
+# ---------------------------------------------------------------------------
+
+
+class TestAskEndpoint:
+    def test_ask_empty_question(self, client):
+        response = client.post("/api/ask", json={"question": ""})
+        assert response.status_code == 400
+
+    def test_ask_no_events(self, client):
+        response = client.post("/api/ask", json={"question": "What happened to device NONEXISTENT in the last 3 days?"})
+        assert response.status_code == 200
+        data = response.json()
+        assert data["answer"] != ""
+        assert data["events"] == []
+        assert data["llm_used"] is False
+        assert data["query_info"]["entity"] == "NONEXISTENT"
+
+    def test_ask_with_events_fallback(self, client, mock_events_collection):
+        now = datetime.now(UTC)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-ask-1",
+                "timestamp": now.isoformat(),
+                "service": "Device",
+                "operation": "Update device",
+                "result": "success",
+                "actor_display": "Admin Bob",
+                "actor_upn": "bob@example.com",
+                "target_displays": ["LAPTOP-001"],
+                "display_summary": "Update device | device: LAPTOP-001 by Admin Bob",
+                "raw_text": "LAPTOP-001 something",
+            }
+        )
+        response = client.post("/api/ask", json={"question": "What happened to device LAPTOP-001 in the last 3 days?"})
+        assert response.status_code == 200
+        data = response.json()
+        assert data["llm_used"] is False
+        assert len(data["events"]) == 1
+        assert data["events"][0]["id"] == "evt-ask-1"
+        assert "LAPTOP-001" in data["answer"]
+        assert data["query_info"]["entity"] == "LAPTOP-001"
+        assert data["query_info"]["event_count"] == 1
+
+    def test_ask_defaults_to_7_days_when_no_time(self, client, mock_events_collection):
+        # Insert an event from 5 days ago
+        five_days_ago = datetime.now(UTC) - timedelta(days=5)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-ask-old",
+                "timestamp": five_days_ago.isoformat(),
+                "service": "Directory",
+                "operation": "Add user",
+                "result": "success",
+                "actor_display": "Alice",
+                "target_displays": ["DESKTOP-999"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+        response = client.post("/api/ask", json={"question": "What happened to DESKTOP-999?"})
+        assert response.status_code == 200
+        data = response.json()
+        assert data["query_info"]["event_count"] == 1
+        assert data["events"][0]["id"] == "evt-ask-old"
+
+    def test_ask_event_outside_time_window(self, client, mock_events_collection):
+        # Event from 10 days ago — outside default 7-day window
+        old = datetime.now(UTC) - timedelta(days=10)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-too-old",
+                "timestamp": old.isoformat(),
+                "service": "Directory",
+                "operation": "Add user",
+                "result": "success",
+                "actor_display": "Alice",
+                "target_displays": ["OLD-DEVICE"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+        response = client.post("/api/ask", json={"question": "What happened to OLD-DEVICE?"})
+        assert response.status_code == 200
+        data = response.json()
+        # Default is 7 days, so 10-day-old event should not match
+        assert data["query_info"]["event_count"] == 0
+
+    def test_ask_with_llm(self, client, mock_events_collection, monkeypatch):
+        now = datetime.now(UTC)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-llm",
+                "timestamp": now.isoformat(),
+                "service": "Device",
+                "operation": "Wipe device",
+                "result": "failure",
+                "actor_display": "System",
+                "target_displays": ["PHONE-001"],
+                "display_summary": "Wipe device | device: PHONE-001 by System",
+                "raw_text": "PHONE-001 wipe failed",
+            }
+        )
+
+        async def fake_llm(question, events, total=None):
+            return "The device had a failed wipe attempt."
+
+        monkeypatch.setattr("routes.ask.LLM_API_KEY", "fake-key")
+        monkeypatch.setattr("routes.ask._call_llm", fake_llm)
+
+        response = client.post("/api/ask", json={"question": "What happened to PHONE-001 in the last day?"})
+        assert response.status_code == 200
+        data = response.json()
+        assert data["llm_used"] is True
+        assert data["answer"] == "The device had a failed wipe attempt."
+        assert len(data["events"]) == 1
+
+    def test_ask_falls_back_when_llm_errors(self, client, mock_events_collection, monkeypatch):
+        now = datetime.now(UTC)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-fallback",
+                "timestamp": now.isoformat(),
+                "service": "Directory",
+                "operation": "Add user",
+                "result": "success",
+                "actor_display": "Alice",
+                "target_displays": ["USER-001"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+
+        async def failing_llm(question, events, total=None):
+            raise RuntimeError("LLM service down")
+
+        monkeypatch.setattr("routes.ask.LLM_API_KEY", "fake-key")
+        monkeypatch.setattr("routes.ask._call_llm", failing_llm)
+
+        response = client.post("/api/ask", json={"question": "What happened to USER-001?"})
+        assert response.status_code == 200
+        data = response.json()
+        assert data["llm_used"] is False  # Falls back
+        assert len(data["events"]) == 1
+        assert "Found 1 event" in data["answer"]
+
+    def test_ask_with_explicit_filters(self, client, mock_events_collection):
+        now = datetime.now(UTC)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-exchange",
+                "timestamp": now.isoformat(),
+                "service": "Exchange",
+                "operation": "Update",
+                "result": "failure",
+                "actor_display": "Alice",
+                "target_displays": ["LAPTOP-001"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-directory",
+                "timestamp": now.isoformat(),
+                "service": "Directory",
+                "operation": "Add user",
+                "result": "success",
+                "actor_display": "Alice",
+                "target_displays": ["LAPTOP-001"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+        response = client.post(
+            "/api/ask",
+            json={"question": "What happened to LAPTOP-001?", "services": ["Exchange"], "result": "failure"},
+        )
+        assert response.status_code == 200
+        data = response.json()
+        assert data["query_info"]["event_count"] == 1
+        assert data["events"][0]["id"] == "evt-exchange"
+
+    def test_ask_with_explicit_actor_filter(self, client, mock_events_collection):
+        now = datetime.now(UTC)
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-bob",
+                "timestamp": now.isoformat(),
+                "service": "Directory",
+                "operation": "Add user",
+                "result": "success",
+                "actor_display": "Bob",
+                "actor_upn": "bob@example.com",
+                "target_displays": ["USER-001"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+        mock_events_collection.insert_one(
+            {
+                "id": "evt-alice",
+                "timestamp": now.isoformat(),
+                "service": "Directory",
+                "operation": "Remove user",
+                "result": "success",
+                "actor_display": "Alice",
+                "actor_upn": "alice@example.com",
+                "target_displays": ["USER-001"],
+                "display_summary": "summary",
+                "raw_text": "raw",
+            }
+        )
+        response = client.post("/api/ask", json={"question": "What happened to USER-001?", "actor": "bob"})
+        assert response.status_code == 200
+        data = response.json()
+        assert data["query_info"]["event_count"] == 1
+        assert data["events"][0]["id"] == "evt-bob"

docker-compose.prod.yml

@@ -0,0 +1,65 @@
+services:
+  mongo:
+    image: mongo:7
+    container_name: aoc-mongo
+    restart: always
+    # Do NOT expose MongoDB port to the host in production
+    # Only backend can reach it via the internal Docker network
+    environment:
+      MONGO_INITDB_ROOT_USERNAME: ${MONGO_ROOT_USERNAME}
+      MONGO_INITDB_ROOT_PASSWORD: ${MONGO_ROOT_PASSWORD}
+    volumes:
+      - mongo_data:/data/db
+    networks:
+      - aoc-internal
+    healthcheck:
+      test: ["CMD", "mongosh", "--eval", "db.adminCommand('ping')"]
+      interval: 10s
+      timeout: 5s
+      retries: 5
+      start_period: 10s
+
+  backend:
+    image: git.cqre.net/cqrenet/aoc-backend:${AOC_VERSION:-latest}
+    container_name: aoc-backend
+    restart: always
+    env_file:
+      - .env
+    environment:
+      MONGO_URI: mongodb://${MONGO_ROOT_USERNAME}:${MONGO_ROOT_PASSWORD}@mongo:27017/
+    depends_on:
+      mongo:
+        condition: service_healthy
+    networks:
+      - aoc-internal
+    healthcheck:
+      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"]
+      interval: 15s
+      timeout: 5s
+      retries: 3
+      start_period: 10s
+
+  nginx:
+    image: nginx:alpine
+    container_name: aoc-nginx
+    restart: always
+    ports:
+      - "80:80"
+      - "443:443"
+    volumes:
+      - ./nginx/nginx.conf:/etc/nginx/nginx.conf:ro
+      - ./nginx/ssl:/etc/nginx/ssl:ro
+    depends_on:
+      backend:
+        condition: service_healthy
+    networks:
+      - aoc-internal
+      - aoc-public
+
+volumes:
+  mongo_data:
+
+networks:
+  aoc-internal:
+    internal: true
+  aoc-public:

docker-compose.yml

@@ -12,8 +12,9 @@ services:
       - mongo_data:/data/db
 
   backend:
-    # For local development you can switch back to: build: ./backend
-    image: git.cqre.net/cqrenet/aoc-backend:v1.0.3
+    build: ./backend
+    # For production, use the pre-built image instead:
+    # image: git.cqre.net/cqrenet/aoc-backend:v1.2.5
     container_name: aoc-backend
     restart: always
     env_file:

nginx/nginx.conf

@@ -0,0 +1,94 @@
+user nginx;
+worker_processes auto;
+error_log /var/log/nginx/error.log warn;
+pid /var/run/nginx.pid;
+
+events {
+    worker_connections 1024;
+}
+
+http {
+    include /etc/nginx/mime.types;
+    default_type application/octet-stream;
+
+    log_format main '$remote_addr - $remote_user [$time_local] "$request" '
+                    '$status $body_bytes_sent "$http_referer" '
+                    '"$http_user_agent" "$http_x_forwarded_for"';
+
+    access_log /var/log/nginx/access.log main;
+
+    sendfile on;
+    tcp_nopush on;
+    tcp_nodelay on;
+    keepalive_timeout 65;
+    types_hash_max_size 2048;
+
+    # Gzip compression
+    gzip on;
+    gzip_vary on;
+    gzip_proxied any;
+    gzip_comp_level 6;
+    gzip_types text/plain text/css text/xml application/json application/javascript application/rss+xml application/atom+xml image/svg+xml;
+
+    # Security headers
+    add_header X-Frame-Options "SAMEORIGIN" always;
+    add_header X-Content-Type-Options "nosniff" always;
+    add_header X-XSS-Protection "1; mode=block" always;
+    add_header Referrer-Policy "strict-origin-when-cross-origin" always;
+
+    # Upstream backend
+    upstream aoc_backend {
+        server backend:8000;
+    }
+
+    # HTTP → HTTPS redirect (optional; enable once TLS is configured)
+    # server {
+    #     listen 80;
+    #     server_name _;
+    #     return 301 https://$host$request_uri;
+    # }
+
+    server {
+        listen 80;
+        server_name _;
+
+        client_max_body_size 50M;
+        proxy_connect_timeout 60s;
+        proxy_send_timeout 60s;
+        proxy_read_timeout 60s;
+
+        location / {
+            proxy_pass http://aoc_backend;
+            proxy_http_version 1.1;
+            proxy_set_header Host $host;
+            proxy_set_header X-Real-IP $remote_addr;
+            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+            proxy_set_header X-Forwarded-Proto $scheme;
+            proxy_buffering off;
+        }
+    }
+
+    # HTTPS server (uncomment and configure once you have certificates)
+    # server {
+    #     listen 443 ssl http2;
+    #     server_name _;
+    #
+    #     ssl_certificate /etc/nginx/ssl/cert.pem;
+    #     ssl_certificate_key /etc/nginx/ssl/key.pem;
+    #     ssl_protocols TLSv1.2 TLSv1.3;
+    #     ssl_ciphers HIGH:!aNULL:!MD5;
+    #     ssl_prefer_server_ciphers on;
+    #
+    #     client_max_body_size 50M;
+    #
+    #     location / {
+    #         proxy_pass http://aoc_backend;
+    #         proxy_http_version 1.1;
+    #         proxy_set_header Host $host;
+    #         proxy_set_header X-Real-IP $remote_addr;
+    #         proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+    #         proxy_set_header X-Forwarded-Proto $scheme;
+    #         proxy_buffering off;
+    #     }
+    # }
+}