feat: playbook agents

* playbook run listener support playbook_run as the parent not just notification_send
flanksource · Jan 9, 2025 · d861117 · d861117
1 parent 43cd2bd
commit d861117
Show file tree

Hide file tree

Showing 20 changed files with 599 additions and 76 deletions.
diff --git a/api/v1/playbook_actions.go b/api/v1/playbook_actions.go
@@ -385,6 +385,8 @@ type AIAction struct {
 	// When enabled, the prompt is simply saved without passing it on to the LLM.
 	DryRun bool `json:"dryRun,omitempty"`
 
+	PlaybookAgents []types.ResourceSelector `json:"playbookAgents,omitempty"`
+
 	// Specify selectors for playbooks. The LLM will recommend the best suited playbooks
 	// in response to the prompt.
 	RecommendPlaybooks []types.ResourceSelector `json:"recommendPlaybooks,omitempty"`

diff --git a/api/v1/zz_generated.deepcopy.go b/api/v1/zz_generated.deepcopy.go
diff --git a/config/crds/mission-control.flanksource.com_playbooks.yaml b/config/crds/mission-control.flanksource.com_playbooks.yaml
@@ -140,6 +140,66 @@ spec:
                             Model name based on the backend chosen.
                             Example: gpt-4o for openai, claude-3-5-sonnet-latest for Anthropic, llama3.1:8b for Ollama
                           type: string
+                        playbookAgents:
+                          items:
+                            properties:
+                              agent:
+                                description: |-
+                                  Agent can be the agent id or the name of the agent.
+                                   Additionally, the special "self" value can be used to select resources without an agent.
+                                type: string
+                              cache:
+                                description: |-
+                                  Cache directives
+                                   'no-cache' (should not fetch from cache but can be cached)
+                                   'no-store' (should not cache)
+                                   'max-age=X' (cache for X duration)
+                                type: string
+                              fieldSelector:
+                                type: string
+                              health:
+                                description: |-
+                                  Health filters resources by the health.
+                                  Multiple healths can be provided separated by comma.
+                                type: string
+                              id:
+                                type: string
+                              includeDeleted:
+                                type: boolean
+                              labelSelector:
+                                type: string
+                              limit:
+                                type: integer
+                              name:
+                                type: string
+                              namespace:
+                                type: string
+                              scope:
+                                description: |-
+                                  Scope is the reference for parent of the resource to select.
+                                  For config items, the scope is the scraper id
+                                  For checks, it's canaries and
+                                  For components, it's topology.
+                                  It can either be a uuid or namespace/name
+                                type: string
+                              search:
+                                description: Search query that applies to the resource
+                                  name, tag & labels.
+                                type: string
+                              statuses:
+                                description: Statuses filter resources by the status
+                                items:
+                                  type: string
+                                type: array
+                              tagSelector:
+                                type: string
+                              types:
+                                description: Types filter resources by the type
+                                items:
+                                  type: string
+                                type: array
+                            type: object
+                          type: array
                         prompt:
                           description: Prompt is the human prompt
                           type: string

diff --git a/config/schemas/playbook-spec.schema.json b/config/schemas/playbook-spec.schema.json
@@ -38,6 +38,12 @@
         "dryRun": {
           "type": "boolean"
         },
+        "playbookAgents": {
+          "items": {
+            "$ref": "#/$defs/ResourceSelector"
+          },
+          "type": "array"
+        },
         "recommendPlaybooks": {
           "items": {
             "$ref": "#/$defs/ResourceSelector"

diff --git a/config/schemas/playbook.schema.json b/config/schemas/playbook.schema.json
@@ -38,6 +38,12 @@
         "dryRun": {
           "type": "boolean"
         },
+        "playbookAgents": {
+          "items": {
+            "$ref": "#/$defs/ResourceSelector"
+          },
+          "type": "array"
+        },
         "recommendPlaybooks": {
           "items": {
             "$ref": "#/$defs/ResourceSelector"

diff --git a/events/event_queue.go b/events/event_queue.go
@@ -12,6 +12,10 @@ import (
 	"gorm.io/gorm/clause"
 )
 
+// events received on this channel are saved to DB.
+// NOTE: Not sure about this one. will probably remove it.
+var EventChan = make(chan models.Event)
+
 type asyncHandlerData struct {
 	fn           func(ctx context.Context, e models.Events) models.Events
 	batchSize    int
@@ -67,6 +71,15 @@ func StartConsumers(ctx context.Context) {
 	for _, fn := range registers {
 		fn(ctx)
 	}
+
+	go func() {
+		for e := range EventChan {
+			if err := ctx.DB().Create(&e).Error; err != nil {
+				log.Errorf("failed to create event: %w", err)
+			}
+		}
+	}()
+
 	// We listen to all PG Notifications on one channel and distribute it to other consumers
 	// based on the events.
 	notifyRouter := pg.NewNotifyRouter()

diff --git a/fixtures/playbooks/ai-agents.yaml b/fixtures/playbooks/ai-agents.yaml
@@ -0,0 +1,47 @@
+---
+# yaml-language-server: $schema=../../config/schemas/playbook.schema.json
+apiVersion: mission-control.flanksource.com/v1
+kind: Playbook
+metadata:
+  name: use-agents
+spec:
+  description: Use AI to diagnose unhealthy config items and send a notification slack
+  category: AI
+  configs:
+    - name: '*'
+  parameters:
+    - name: prompt
+      label: Prompt
+      default: Find out why {{.config.name}} is unhealthy and report in one short sentence.
+      properties:
+        multiline: 'true'
+  actions:
+    - name: query LLM
+      ai:
+        connection: connection://mc/anthropic
+        playbookAgents:
+          - namespace: mc
+        systemPrompt: |
+          **Role:** Seasoned Kubernetes engineer and Diagnostic Expert
+
+          **Objective:** Assist users in diagnosing issues with unhealthy Kubernetes resources by analyzing provided manifests and related resources.
+
+          **Instructions:**
+
+          1. **Resource Analysis:** Examine the manifest of the unhealthy resource thoroughly.
+          2. **Contextual Investigation:** Consider additional related resources provided (e.g., pods, replica sets, namespaces) to gain a comprehensive understanding of the issue.
+          4. **One-Time Diagnosis:** Aim to diagnose the issue in a single response without requiring follow-up questions.
+        prompt: '{{.params.prompt}}'
+        changes:
+          since: 2d
+        analysis:
+          since: 2d
+        relationships:
+          - depth: 3
+            direction: outgoing
+            changes:
+              since: 24h
+          - depth: 5
+            direction: incoming
+            changes:
+              since: 24h
diff --git a/fixtures/playbooks/ai-diagnose-aws-ollama.yaml b/fixtures/playbooks/ai-diagnose-aws-ollama.yaml
@@ -6,6 +6,7 @@ metadata:
   name: diagnose-aws-resource
 spec:
   description: Use AI to diagnose AWS resources
+  category: AI
   configs:
     - types:
         - AWS::CloudFormation::Stack

diff --git a/fixtures/playbooks/ai-diagnose-kubernetes-resource.yaml b/fixtures/playbooks/ai-diagnose-kubernetes-resource.yaml
@@ -6,6 +6,7 @@ metadata:
   name: diagnose-kubernetes-resource
 spec:
   description: Use AI to diagnose unhealthy kubernetes resources
+  category: AI
   configs:
     - types:
         - Kubernetes::DaemonSet

diff --git a/fixtures/playbooks/ai-diagnose-slack-notification.yaml b/fixtures/playbooks/ai-diagnose-slack-notification.yaml
@@ -6,8 +6,9 @@ metadata:
   name: diagnose-and-send-to-slack
 spec:
   description: Use AI to diagnose unhealthy config items and send a notification slack
+  category: AI
   configs:
-    - health: unhealthy,warning
+    - health: unhealthy,warning,unknown,healthy
   parameters:
     - name: prompt
       label: Prompt

diff --git a/fixtures/playbooks/ai-recommend-playbook.yaml b/fixtures/playbooks/ai-recommend-playbook.yaml
@@ -7,14 +7,8 @@ metadata:
 spec:
   description: Use AI to diagnose unhealthy config items and send a notification slack
   configs:
-<<<<<<< HEAD
-    - health: unhealthy,warning
-||||||| parent of 4e54e54 (feat(ai playbook action): formats support & playbook recommender)
-=======
-    - healths:
-        - unhealthy
-        - warning
->>>>>>> 4e54e54 (feat(ai playbook action): formats support & playbook recommender)
+    - health: unhealthy,warning,unknown,healthy
+  category: AI
   parameters:
     - name: prompt
       label: Prompt

diff --git a/go.mod b/go.mod
@@ -14,7 +14,7 @@ require (
 	github.com/containrrr/shoutrrr v0.8.0
 	github.com/fergusstrange/embedded-postgres v1.29.0 // indirect
 	github.com/flanksource/commons v1.35.3
-	github.com/flanksource/duty v1.0.800
+	github.com/flanksource/duty v1.0.802
 	github.com/flanksource/gomplate/v3 v3.24.50
 	github.com/flanksource/kopper v1.0.11
 	github.com/gomarkdown/markdown v0.0.0-20240419095408-642f0ee99ae2
@@ -92,6 +92,7 @@ require (
 	dario.cat/mergo v1.0.0 // indirect
 	github.com/GehirnInc/crypt v0.0.0-20200316065508-bb7000b8a962 // indirect
 	github.com/Masterminds/semver/v3 v3.3.1 // indirect
+	github.com/Masterminds/sprig/v3 v3.2.3 // indirect
 	github.com/Microsoft/go-winio v0.6.1 // indirect
 	github.com/ProtonMail/go-crypto v1.0.0 // indirect
 	github.com/RaveNoX/go-jsonmerge v1.0.0 // indirect
@@ -167,13 +168,15 @@ require (
 	github.com/google/pprof v0.0.0-20240727154555-813a5fbdbec8 // indirect
 	github.com/google/s2a-go v0.1.8 // indirect
 	github.com/googleapis/enterprise-certificate-proxy v0.3.4 // indirect
+	github.com/goph/emperror v0.17.2 // indirect
 	github.com/gorilla/websocket v1.5.1 // indirect
 	github.com/gosimple/unidecode v1.0.1 // indirect
 	github.com/gregjones/httpcache v0.0.0-20190611155906-901d90724c79 // indirect
 	github.com/grpc-ecosystem/grpc-gateway/v2 v2.20.0 // indirect
 	github.com/hairyhenderson/yaml v0.0.0-20220618171115-2d35fca545ce // indirect
 	github.com/hashicorp/hcl/v2 v2.21.0 // indirect
 	github.com/hirochachacha/go-smb2 v1.1.0 // indirect
+	github.com/huandu/xstrings v1.3.3 // indirect
 	github.com/invopop/jsonschema v0.12.0 // indirect
 	github.com/itchyny/gojq v0.12.17 // indirect
 	github.com/itchyny/timefmt-go v0.1.6 // indirect
@@ -204,10 +207,12 @@ require (
 	github.com/moby/spdystream v0.4.0 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
 	github.com/mxk/go-flowrate v0.0.0-20140419014527-cca7078d478f // indirect
+	github.com/nikolalohinski/gonja v1.5.3 // indirect
 	github.com/ohler55/ojg v1.25.0 // indirect
 	github.com/oklog/ulid/v2 v2.1.0 // indirect
 	github.com/opencontainers/go-digest v1.0.0 // indirect
 	github.com/orcaman/concurrent-map/v2 v2.0.1 // indirect
+	github.com/pelletier/go-toml/v2 v2.0.9 // indirect
 	github.com/peterbourgon/diskv v2.0.1+incompatible // indirect
 	github.com/pjbgf/sha1cd v0.3.0 // indirect
 	github.com/pkg/sftp v1.13.6 // indirect
@@ -223,10 +228,12 @@ require (
 	github.com/sergi/go-diff v1.3.2-0.20230802210424-5b0b94c5c0d3 // indirect
 	github.com/shirou/gopsutil/v3 v3.24.5 // indirect
 	github.com/shoenig/go-m1cpu v0.1.6 // indirect
+	github.com/shopspring/decimal v1.2.0 // indirect
 	github.com/shurcooL/githubv4 v0.0.0-20190718010115-4ba037080260 // indirect
 	github.com/shurcooL/graphql v0.0.0-20181231061246-d48a9a75455f // indirect
 	github.com/sirupsen/logrus v1.9.3 // indirect
 	github.com/skeema/knownhosts v1.2.2 // indirect
+	github.com/spf13/cast v1.5.0 // indirect
 	github.com/stoewer/go-strcase v1.3.0 // indirect
 	github.com/tidwall/gjson v1.17.0 // indirect
 	github.com/tidwall/match v1.1.1 // indirect
@@ -243,6 +250,7 @@ require (
 	github.com/xeipuuv/gojsonpointer v0.0.0-20180127040702-4e3ac2762d5f // indirect
 	github.com/xeipuuv/gojsonreference v0.0.0-20180127040603-bd5ef7bd5415 // indirect
 	github.com/xi2/xz v0.0.0-20171230120015-48954b6210f8 // indirect
+	github.com/yargevad/filepathx v1.0.0 // indirect
 	github.com/yuin/gopher-lua v1.1.1 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // indirect
 	github.com/zclconf/go-cty v1.15.0 // indirect
@@ -251,6 +259,7 @@ require (
 	go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.22.0 // indirect
 	go.opentelemetry.io/otel/metric v1.29.0 // indirect
 	go.opentelemetry.io/proto/otlp v1.3.1 // indirect
+	go.starlark.net v0.0.0-20231121155337-90ade8b19d09 // indirect
 	golang.org/x/exp v0.0.0-20241108190413-2d47ceb2692f // indirect
 	golang.org/x/mod v0.22.0 // indirect
 	golang.org/x/sync v0.10.0 // indirect