@vfarcic/dot-ai 0.192.0 → 0.193.0

package/prompts/partials/visualization-output.md

@@ -0,0 +1,93 @@
+## Output Format
+
+When you have gathered sufficient information, respond with ONLY a JSON object (no markdown code fences, no extra text):
+
+{
+  "title": "Descriptive title based on what was found",
+  "visualizations": [...],
+  "insights": [...]
+}
+
+**If you cannot fulfill the user's request**, return (still as raw JSON, no code fences):
+
+{
+  "title": "Unable to Perform Operation",
+  "visualizations": [],
+  "insights": ["Explanation of why the request cannot be fulfilled and what tool or action should be used instead."],
+  "status": "error"
+}
+
+## Visualization Types
+
+Generate as many or as few visualizations as add value. You can:
+- Include multiple visualizations of the same type
+- Skip any type entirely if it doesn't add value for this data
+- Use whatever combination best represents what you found
+
+Each visualization:
+```
+{
+  "id": "unique-id",
+  "label": "Tab Label",
+  "type": "mermaid" | "table" | "cards" | "code" | "bar-chart",
+  "content": <type-specific-content>
+}
+```
+
+### mermaid
+- `content`: Valid Mermaid diagram string
+- Use `graph TD` or `graph LR` depending on relationship type
+- Use subgraphs to group related items
+- Arrows: `-->` for direct relationships, `-.->` for indirect/inferred
+- **Styling rules**:
+  - Only style nodes that represent errors or problems
+  - For error nodes: `style NodeId fill:#ef4444,stroke:#dc2626,color:#ffffff` (white text on red background)
+  - **CRITICAL: Text contrast** - Always ensure text is readable against the background:
+    - Dark backgrounds (red, gray, etc.) → use `color:#ffffff` (white text)
+    - Light backgrounds → use `color:#ffffff` or `color:#000000` depending on shade
+    - NEVER use similar colors for text and background (e.g., red text on pink background)
+  - Do NOT add colors or styling to healthy/normal nodes - use default Mermaid styling
+  - Truncate UUIDs to first 8 characters (e.g., `pvc-508555a4...`)
+  - Keep node labels under 30 characters when possible
+
+### table
+- `content`: `{ "headers": ["Col1", "Col2"], "rows": [["val1", "val2"]], "rowStatuses": ["error", null, "ok"] }`
+- Choose columns relevant to the data present
+- Include status/condition information when available
+- Optional `rowStatuses` array: Indicate row-level status. Values: `"error"`, `"warning"`, `"ok"`, or `null`. Array length should match rows.
+
+### cards
+- `content`: `[{ "id": "unique", "title": "Name", "subtitle": "Type or status", "description": "Details", "tags": ["tag1"], "status": "error" }]`
+- Use for items where individual status or comparison matters
+- Tags should reflect actual state from the data
+- Optional `status` field: Indicate item status. Values: `"error"`, `"warning"`, `"ok"`. Omit for neutral items.
+
+### code
+- `content`: `{ "language": "yaml" | "json" | "bash", "code": "..." }`
+- Use sparingly - only when raw output adds value
+
+### bar-chart
+- `content`: `{ "data": [{ "label": "...", "value": 8.5, "max": 10, "status": "ok" }], "unit": "Gi", "orientation": "horizontal" }`
+- `data[].label`: Category label
+- `data[].value`: Numeric value
+- `data[].max`: Optional maximum for percentage calculation
+- `data[].status`: Optional `"error"`, `"warning"`, `"ok"`. Omit for neutral items.
+- `unit`: Optional unit label
+- `orientation`: Optional `"horizontal"` (default) or `"vertical"`
+
+## Insights
+
+Generate insights that add value beyond what someone could see by just reading the raw data. Prioritize non-obvious findings over summaries.
+
+Each insight should:
+- Reference specific items from the data
+- Explain WHY it matters, not just WHAT you found
+- Be actionable when highlighting issues
+
+## Rules
+
+1. **Data-driven only** - Base visualizations on actual data present
+2. **Skip empty visualizations** - Don't include visualizations with no meaningful content
+3. **Valid output** - Ensure Mermaid syntax is correct and JSON is valid
+4. **JSON only** - No markdown fences, no explanations outside the JSON structure
+5. **Use validate_mermaid** - Before returning any Mermaid diagrams, validate them using the validate_mermaid tool

package/prompts/query-system.md

@@ -1,6 +1,12 @@
 # Kubernetes Cluster Query Agent
 
-You are a Kubernetes cluster analyst. Use the available tools to answer the user's query.
+You are a read-only Kubernetes cluster analyst. Use the available tools to answer questions about cluster state, resources, and capabilities.
+
+## Critical Constraint: Read-Only Operations
+
+You can ONLY query information. You CANNOT modify the cluster (scale, create, update, delete, restart, patch, rollback, etc.).
+
+If the user requests a modification, immediately return a response (following the output format below) explaining that you can only query cluster information and they should use the `operate` tool instead. Do not investigate or ask clarifying questions about modification requests.
 
 ## Thoroughness
 
@@ -8,14 +14,4 @@ You are a Kubernetes cluster analyst. Use the available tools to answer the user
 - For conceptual queries, check ALL resource kinds returned by capability search
 - When in doubt, search more rather than less
 
-## Output Format
-
-When you have gathered sufficient information, respond with ONLY this JSON:
-
-```json
-{
-  "summary": "Human-readable summary of what was found"
-}
-```
-
-No text before or after the JSON.
+{{{outputInstructions}}}

package/prompts/resource-selection.md

@@ -86,6 +86,10 @@ Respond with ONLY a JSON object. The format depends on whether capabilities can
 
 Return solutions with resources. Include the `resourceName` field for each resource (extract from Available Resources list).
 
+**Solution type field (REQUIRED):**
+- `"type": "single"` - when the solution uses exactly ONE resource
+- `"type": "combination"` - when the solution uses TWO OR MORE resources together
+
 ```json
 {
   "solutions": [

package/prompts/visualize.md

@@ -16,71 +16,4 @@ You are a Kubernetes visualization expert. Analyze the provided data and generat
 
 If the provided data isn't sufficient for deep analysis, **actively use tools** to gather additional information. Don't just note uncertainties - investigate them.
 
-## Output Format
-
-Respond with ONLY a JSON object (no markdown code fences, no extra text):
-
-{
-  "title": "Descriptive title based on what was found",
-  "visualizations": [...],
-  "insights": [...]
-}
-
-## Visualization Types
-
-Generate as many or as few visualizations as add value. You can:
-- Include multiple visualizations of the same type
-- Skip any type entirely if it doesn't add value for this data
-- Use whatever combination best represents what you found
-
-Each visualization:
-```
-{
-  "id": "unique-id",
-  "label": "Tab Label",
-  "type": "mermaid" | "table" | "cards" | "code",
-  "content": <type-specific-content>
-}
-```
-
-### mermaid
-- `content`: Valid Mermaid diagram string
-- Use `graph TD` or `graph LR` depending on relationship type
-- Use subgraphs to group related items
-- Arrows: `-->` for direct relationships, `-.->` for indirect/inferred
-- **Syntax rules**:
-  - Node names in `class` statements MUST be comma-separated: `class a,b,c style` (not space-separated)
-  - If using `classDef`, ALWAYS specify both `fill` AND `color` (text) with sufficient contrast
-  - Truncate UUIDs to first 8 characters (e.g., `pvc-508555a4...`)
-  - Keep node labels under 30 characters when possible
-
-### table
-- `content`: `{ "headers": ["Col1", "Col2"], "rows": [["val1", "val2"]] }`
-- Choose columns relevant to the data present
-- Include status/condition information when available
-
-### cards
-- `content`: `[{ "id": "unique", "title": "Name", "subtitle": "Type or status", "description": "Details", "tags": ["tag1"] }]`
-- Use for items where individual status or comparison matters
-- Tags should reflect actual state from the data
-
-### code
-- `content`: `{ "language": "yaml" | "json" | "bash", "code": "..." }`
-- Use sparingly - only when raw output adds value
-
-## Insights
-
-Generate insights that add value beyond what someone could see by just reading the raw data. Prioritize non-obvious findings over summaries.
-
-Each insight should:
-- Reference specific items from the data
-- Explain WHY it matters, not just WHAT you found
-- Be actionable when highlighting issues
-
-## Rules
-
-1. **Data-driven only** - Base visualizations on actual data present
-2. **Skip empty visualizations** - Don't include visualizations with no meaningful content
-3. **Valid output** - Ensure Mermaid syntax is correct and JSON is valid
-4. **JSON only** - No markdown fences, no explanations outside the JSON structure
-5. **Use validate_mermaid** - Before returning any Mermaid diagrams, validate them using the validate_mermaid tool
+{{{visualizationOutput}}}

package/scripts/reset-sync-and-scan-test-cluster.sh

@@ -0,0 +1,120 @@
+#!/bin/bash
+
+# Reset ResourceSyncConfig and CapabilityScanConfig
+# This script:
+# 1. Deletes ResourceSyncConfig and CapabilityScanConfig CRs from the cluster
+# 2. Deletes all capabilities data via MCP API
+# 3. Re-applies the CRs in the dot-ai namespace
+#
+# Usage:
+#   ./scripts/reset-sync-and-scan-test-cluster.sh
+
+set -e
+
+# Configuration - defaults match test cluster settings
+KUBECONFIG="${KUBECONFIG:-./kubeconfig-test.yaml}"
+export KUBECONFIG
+
+NAMESPACE="${NAMESPACE:-dot-ai}"
+MCP_URL="${MCP_URL:-http://dot-ai.127.0.0.1.nip.io:8180}"
+AUTH_TOKEN="${AUTH_TOKEN:-test-auth-token-integration}"
+RESOURCE_SYNC_NAME="${RESOURCE_SYNC_NAME:-default-sync}"
+CAPABILITY_SCAN_NAME="${CAPABILITY_SCAN_NAME:-default-scan}"
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+echo -e "${YELLOW}=== Reset ResourceSyncConfig and CapabilityScanConfig ===${NC}"
+echo "Kubeconfig: $KUBECONFIG"
+echo "Namespace: $NAMESPACE"
+echo "MCP URL: $MCP_URL"
+echo ""
+
+# Step 1: Delete existing CRs
+echo -e "${YELLOW}Step 1: Deleting existing CRs...${NC}"
+
+if kubectl get resourcesyncconfig "$RESOURCE_SYNC_NAME" -n "$NAMESPACE" &>/dev/null; then
+    echo "Deleting ResourceSyncConfig/$RESOURCE_SYNC_NAME..."
+    kubectl delete resourcesyncconfig "$RESOURCE_SYNC_NAME" -n "$NAMESPACE" --ignore-not-found
+    echo -e "${GREEN}ResourceSyncConfig deleted${NC}"
+else
+    echo "ResourceSyncConfig/$RESOURCE_SYNC_NAME not found, skipping..."
+fi
+
+if kubectl get capabilityscanconfig "$CAPABILITY_SCAN_NAME" -n "$NAMESPACE" &>/dev/null; then
+    echo "Deleting CapabilityScanConfig/$CAPABILITY_SCAN_NAME..."
+    kubectl delete capabilityscanconfig "$CAPABILITY_SCAN_NAME" -n "$NAMESPACE" --ignore-not-found
+    echo -e "${GREEN}CapabilityScanConfig deleted${NC}"
+else
+    echo "CapabilityScanConfig/$CAPABILITY_SCAN_NAME not found, skipping..."
+fi
+
+echo ""
+
+# Step 2: Delete all capabilities via MCP API
+echo -e "${YELLOW}Step 2: Deleting all capabilities via MCP API...${NC}"
+
+response=$(curl -s -X POST "$MCP_URL/api/v1/tools/manageOrgData" \
+    -H "Content-Type: application/json" \
+    -H "Authorization: Bearer $AUTH_TOKEN" \
+    -d '{"dataType": "capabilities", "operation": "deleteAll"}')
+
+if echo "$response" | grep -q '"success":true'; then
+    echo -e "${GREEN}All capabilities deleted${NC}"
+else
+    echo -e "${RED}Failed to delete capabilities: $response${NC}"
+fi
+
+echo ""
+
+# Step 3: Apply the CRs
+echo -e "${YELLOW}Step 3: Applying CRs to namespace $NAMESPACE...${NC}"
+
+# Create ResourceSyncConfig
+echo "Creating ResourceSyncConfig..."
+cat <<EOF | kubectl apply -f -
+apiVersion: dot-ai.devopstoolkit.live/v1alpha1
+kind: ResourceSyncConfig
+metadata:
+  name: $RESOURCE_SYNC_NAME
+  namespace: $NAMESPACE
+spec:
+  mcpEndpoint: http://dot-ai.$NAMESPACE.svc.cluster.local:3456/api/v1/resources/sync
+  mcpAuthSecretRef:
+    name: dot-ai-secrets
+    key: auth-token
+  debounceWindowSeconds: 10
+  resyncIntervalMinutes: 60
+EOF
+echo -e "${GREEN}ResourceSyncConfig created${NC}"
+
+# Create CapabilityScanConfig
+echo "Creating CapabilityScanConfig..."
+cat <<EOF | kubectl apply -f -
+apiVersion: dot-ai.devopstoolkit.live/v1alpha1
+kind: CapabilityScanConfig
+metadata:
+  name: $CAPABILITY_SCAN_NAME
+  namespace: $NAMESPACE
+spec:
+  mcp:
+    endpoint: http://dot-ai.$NAMESPACE.svc.cluster.local:3456/api/v1/tools/manageOrgData
+    authSecretRef:
+      name: dot-ai-secrets
+      key: auth-token
+EOF
+echo -e "${GREEN}CapabilityScanConfig created${NC}"
+
+echo ""
+echo -e "${GREEN}=== Reset complete ===${NC}"
+echo ""
+echo "The controller will now:"
+echo "  - Start syncing resources to the 'resources' collection"
+echo "  - Start scanning capabilities to the 'capabilities' collection"
+echo ""
+echo "Monitor progress with:"
+echo "  kubectl get resourcesyncconfig $RESOURCE_SYNC_NAME -n $NAMESPACE -o yaml"
+echo "  kubectl get capabilityscanconfig $CAPABILITY_SCAN_NAME -n $NAMESPACE -o yaml"

package/shared-prompts/prd-create.md

@@ -11,7 +11,7 @@ category: project-management
 You are helping create a Product Requirements Document (PRD) for a new feature. This process involves two main components:
 
 1. **GitHub Issue**: Short, immutable concept description that links to the detailed PRD
-2. **PRD File**: Project management document with milestone tracking, progress logs, and implementation plan
+2. **PRD File**: Project management document with milestone tracking and implementation plan
 
 ## Process