| agent: | Auto Exec |
What is an "Expert"? How do we create our own expert?
Add credentials for various integrations
Managing workspaces and access control
DagKnows Architecture Overview
Setting up SSO via Azure AD for Dagknows
Enable "Auto Exec" and "Send Execution Result to LLM" in "Adjust Settings" if desired
(Optionally) Add ubuntu user to docker group and refresh group membership
Deployment of an EKS Cluster with Worker Nodes in AWS
Adding, Deleting, Listing DagKnows Proxy credentials or key-value pairs
Comprehensive AWS Security and Compliance Evaluation Workflow (SOC2 Super Runbook)
AWS EKS Version Update 1.29 to 1.30 via terraform
Instruction to allow WinRM connection
MSP Usecase: User Onboarding Azure + M365
Post a message to a Slack channel
How to debug a kafka cluster and kafka topics?
Open VPN Troubleshooting (Powershell)
Execute a simple task on the proxy
Assign the proxy role to a user
Create roles to access credentials in proxy
Install OpenVPN client on Windows laptop
Setup Kubernetes kubectl and Minikube on Ubuntu 22.04 LTS
Install Prometheus and Grafana on the minikube cluster on EC2 instance in the monitoring namespace
update the EKS versions in different clusters
AI agent session 2024-09-12T09:36:14-07:00 by Sarang Dharmapurikar
Parse EDN content and give a JSON out
Check whether a user is there on Azure AD and if the user account status is enabled
Get the input parameters of a Jenkins pipeline
QZz2rpIPKUPPKXbKlUpMProcess Grafana Alerts
Process Grafana Alerts
There was a problem that the LLM was not able to address. Please rephrase your prompt and try again.
Process Grafana Alert
inputs
outputs
problem = payload["alerts"][0]["annotations"]["description"]
#{"alerts":[{"annotations":{"description":"Endpoint is Down: http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu in dagknows namespace.","summary":"Endpoint Down."},"dashboardURL":"","endsAt":"0001-01-01T00:00:00Z","fingerprint":"8b773b64aaeebd15","generatorURL":"http://localhost:3000/alerting/grafana/ee9l72z18oikgf/view?orgId=1","labels":{"alertname":"ProbeFailing-MinikubeApplication (copy)","grafana_folder":"alert_rule_folder_1m","instance":"http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu","job":"blackbox-exporter-metrics"},"panelURL":"","silenceURL":"http://localhost:3000/alerting/silence/new?alertmanager=grafana&matcher=alertname%3DProbeFailing-MinikubeApplication+%28copy%29&matcher=grafana_folder%3Dalert_rule_folder_1m&matcher=instance%3Dhttp%3A%2F%2Fdemo.dagknows.com%3A8081%2Fapi%2Ftasks%2FZ6ylcZTzO1xURkNUngRu&matcher=job%3Dblackbox-exporter-metrics&orgId=1","startsAt":"2025-06-02T01:59:20Z","status":"firing","valueString":"[ var='A' labels={instance=http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu, job=blackbox-exporter-metrics} value=0 ], [ var='C' labels={instance=http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu, job=blackbox-exporter-metrics} value=1 ]","values":{"A":0,"C":1}}],"commonAnnotations":{"description":"Endpoint is Down: http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu in dagknows namespace.","summary":"Endpoint Down."},"commonLabels":{"alertname":"ProbeFailing-MinikubeApplication (copy)","grafana_folder":"alert_rule_folder_1m","instance":"http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu","job":"blackbox-exporter-metrics"},"externalURL":"http://localhost:3000/","groupKey":"{}/{__grafana_autogenerated__=\"true\"}/{__grafana_receiver__=\"dk_webhook_security_group_demo\"}/{__grafana_route_settings_hash__=\"31cd9e4f79dd0eae\"}:{alertname=\"ProbeFailing-MinikubeApplication (copy)\", grafana_folder=\"alert_rule_folder_1m\"}","groupLabels":{"alertname":"ProbeFailing-MinikubeApplication (copy)","grafana_folder":"alert_rule_folder_1m"},"message":"**Firing**\n\nValue: A=0, C=1\nLabels:\n - alertname = ProbeFailing-MinikubeApplication (copy)\n - grafana_folder = alert_rule_folder_1m\n - instance = http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu\n - job = blackbox-exporter-metrics\nAnnotations:\n - description = Endpoint is Down: http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu in dagknows namespace.\n - summary = Endpoint Down.\nSource: http://localhost:3000/alerting/grafana/ee9l72z18oikgf/view?orgId=1\nSilence: http://localhost:3000/alerting/silence/new?alertmanager=grafana&matcher=alertname%3DProbeFailing-MinikubeApplication+%28copy%29&matcher=grafana_folder%3Dalert_rule_folder_1m&matcher=instance%3Dhttp%3A%2F%2Fdemo.dagknows.com%3A8081%2Fapi%2Ftasks%2FZ6ylcZTzO1xURkNUngRu&matcher=job%3Dblackbox-exporter-metrics&orgId=1\n","orgId":1,"receiver":"dk_webhook_security_group_demo","state":"alerting","status":"firing","title":"[FIRING:1] ProbeFailing-MinikubeApplication (copy) alert_rule_folder_1m (http://demo.dagknows.com:8081/api/tasks/Z6ylcZTzO1xURkNUngRu blackbox-exporter-metrics)","truncatedAlerts":0,"version":"1"}
copied- 1IUkBwsZXofY0OwP4xSAsChoose and execute a runbook for a given problem
1
Choose and execute a runbook for a given problem
There was a problem that the LLM was not able to address. Please rephrase your prompt and try again.inputsoutputs1- 1.1SC7PHq17Ww7QXoJmcdMZSearch runbooks for a given problem
1.1
Search runbooks for a given problem
There was a problem that the LLM was not able to address. Please rephrase your prompt and try again.inputsoutputsimport json import copy cred_label = "dktoken" search_url = f"https://dev.dagknows.com/api/tasks/?with_pending_perms=false&page_key=0&page_size=10&q={problem}&knn.k=3&knn.nc=10&order_by=elastic&tags=tshoot" op = _rest_api("get", search_url, "", {}, "", cred_label) resp = op.json() tmptasks = [] for task in resp["tasks"]: tmptask = copy.deepcopy(task) for key, value in task.items(): if key not in ["title", "description", "input_params", "output_params", "id", "tags"]: tmptask.pop(key, None) tmptasks.append(tmptask) print(json.dumps(tmptasks, indent=4, default=str)) # print(json.dumps(op, indent=4)) relevant_runbooks = tmptaskscopied1.1 - 1.2wIRFKmUodwg7ewunpsmTSend the problem and the relevant runbooks to LLM and get the runbook to execute
1.2
Send the problem and the relevant runbooks to LLM and get the runbook to execute
There was a problem that the LLM was not able to address. Please rephrase your prompt and try again.inputsoutputsimport json import openai # Get the API key from the environment variable api_key = getEnvVar('OPENAI_API_KEY') # Initialize the OpenAI client with the API key openai.api_key = api_key system_prompt = """ You are a DevOps, ITOps, and SRE expert who excells at picking the right tool to solve a problem. User will give you a problem to be solved along with an array of tools available. Your task is to identify the right tool to execute along with the input values required for execution. The list of tools is formatted as follows: [ { "id" : "<A unique ID of the tool>" "title": "<A one line title that tells what this tool does>", "description": " Description of the tool", "input_params": [ # List of input parameters the tool accepts, each parameter is a dictionary { "default_value": "<Optional. Default value of the >", "description": "<Optional. Description of the input parameter>", "name": "<name of the input parameter>", "param_type": "<Python type of the input parameter>", "required": <Boolean indicating if the input value is required or not> } ] "output_params": [ # List of output parameters the tool produces, Each parameter is a dictionary just like input parameter dict. { <Same dictionary as above for an input parameter> } ] }, ] Your output should be strictly a JSON as follows: { "id" : "<ID of the tool to execute. If no suitable tool found then set it to null.>", "input_params" : [ # List of input parameters with values to assign for execution. Make sure required ones are there. { "name" : "<Name of the param to assign value to>", "value" : "<The value to be assigned to the input. Make sure it matches the input type>" } ] } Do not give any preamble or explanation. Just JSON. """ user_prompt = f""" Here's the problem to solve: {problem} Here's the list of available tools: {json.dumps(relevant_runbooks, indent=4)} """ messages = [ {"role" : "system", "content" : system_prompt}, {"role" : "user", "content" : user_prompt} ] response = openai.chat.completions.create( model='gpt-4o', messages=messages, temperature=0 ) raw_content = response.choices[0].message.content cleaned_output = "\n".join( [line for line in raw_content.split("\n") if not line.startswith("```")] ).strip() chosen_tool = json.loads(cleaned_output) print(chosen_tool)copied1.2 - 1.3WfZtepDNEJFpiaRIlCO3Execute the chosen runbook
1.3
There was a problem that the LLM was not able to address. Please rephrase your prompt and try again.inputsoutputscred_label = "dktoken" api_endpoint = f"https://dev.dagknows.com/api/tasks/{chosen_tool["id"]}/execute" param_values = {} for iparam in chosen_tool["input_params"]: name = iparam["name"] value = iparam["value"] param_values[name] = value job = { "proxy_alias": "dev", "param_values": param_values, "special_param_values": { "stop_after_first_failure": "", "schedule_options": { "start": None, "repeat_interval": "", "repeat_interval_type": "", "end_option": "", "end": None, "num_iterations": "" }, "wsid": "", "proxy_alias": "dev", "button": "regular" }, "output_params": {}, "runbook_task_id": chosen_tool["id"], "starting_child_path": "", "conv_id": f"tconv_{chosen_tool["id"]}", "role": "", "experimental": {} } body = {} body["job"] = job print("API endpoint: ", api_endpoint) print("Body: ", json.dumps(body, indent=4)) op = _rest_api("post", api_endpoint, "", {}, body, cred_label).json() print(json.dumps(op, indent=4, default=str)) job_id = op["job"]["job_id"] job_url = f"https://dev.dagknows.com/tasks/{chosen_tool['id']}?job_id={job_id}&iter=0" task_title = op["job"]["title"] table = context.newtable() # This creates a new table table.num_rows = 2 table.num_cols = 4 table.title = "Chosen runbook" table.has_header_row = True # If the first row is header row then set to True else False table.setval(0, 0, "ID") table.setval(0, 1, "Title") table.setval(0, 2, "Params") table.setval(0, 3, "Execution results") table.setval(1, 0, chosen_tool["id"]) table.setval(1, 1, task_title) table.setval(1, 2, json.dumps(param_values, indent=4, default=str)) table.setval(1, 3, f"<a href={job_url}>Link</a>")copied1.3