main
md 182 lines 5.45 KB
Rendered Raw
1 ---
2 title: CoPilot actions
3 description: Run repeatable response actions across endpoints using CoPilot + Velociraptor, and visualize results in Grafana.
4 ---
5
6 # CoPilot actions
7
8 **Menu:** Agents → CoPilot Actions
9
10 CoPilot Actions provides a more flexible way to launch endpoint actions (response + collection) across your infrastructure.
11
12 At a high level it combines:
13 - **CoPilot** (operator UI)
14 - **Velociraptor** (executes artifacts to run the action)
15 - **Custom scripts** (the action logic)
16 - **Grafana** (dashboards to visualize action results)
17
18 ![CoPilot Actions](../../assets/ui/agents-copilot-actions.png)
19
20 Repo (required for setup assets + implementation details):
21 - https://github.com/socfortress/CoPilot-Action
22
23 ---
24
25 ## Why this exists (vs “traditional Wazuh Active Response”)
26
27 Wazuh Active Response is powerful, but at scale it can become cumbersome to:
28 - deploy scripts to endpoints
29 - keep action logic consistent across OSes
30 - manage parameters and rollouts cleanly
31
32 CoPilot Actions is designed to make automated responses **simpler to run and easier to operationalize**.
33
34 ---
35
36 ## How it works
37
38 Conceptually:
39
40 1) You choose an action in CoPilot
41 2) CoPilot invokes a **Velociraptor artifact** (Windows or Linux)
42 3) The artifact downloads/executes the **action script** for the selected action
43 4) Results are written to logs and ingested into the stack
44 5) Grafana dashboards let you explore outcomes over time
45
46 CoPilot’s own UI also shows:
47 - supported OS/technology
48 - action metadata + description
49 - a link to the underlying source code repo for the action
50
51 ![Action details (placeholder)](../../assets/ui/agents-copilot-actions-action-details.png)
52
53 ---
54
55 ## Prerequisites
56
57 - CoPilot is deployed and you can access **Agents → CoPilot Actions**.
58 - Velociraptor server + clients are deployed (the CoPilot-Action repo recommends Velociraptor **0.74.1+**).
59 - CoPilot can authenticate to Velociraptor (Connector configured).
60
61 ---
62
63 ## Setup checklist (recommended)
64
65 This is the practical “get it working end-to-end” checklist.
66
67 ### 1) Import the required Velociraptor artifacts
68
69 From the repo:
70 - `velociraptor/Linux.Execute.RemoteBashScript.yaml`
71 - `velociraptor/Windows.Execute.RemotePowerShellScript.yaml`
72
73 Import them into Velociraptor (UI):
74 - **View Artifacts → Upload Artifacts**
75
76 Why: these artifacts are the execution layer that downloads and runs action scripts consistently.
77
78 ### 2) Confirm the CoPilot ↔ Velociraptor connector
79
80 In CoPilot, configure and test the Velociraptor connector:
81 - **Connectors → Velociraptor**
82
83 Why: CoPilot needs API access to launch the artifact executions.
84
85 ### 3) Ensure endpoints can download action scripts
86
87 CoPilot Actions commonly downloads scripts from public GitHub repos.
88
89 Verify:
90 - endpoints have egress to `raw.githubusercontent.com` (or wherever your scripts live)
91 - DNS + TLS inspection/proxy rules won’t block downloads
92
93 ### 4) (Windows) Verify PowerShell can run scripts
94
95 If a Windows action relies on PowerShell, validate the execution policy and permissions on target endpoints.
96
97 (Example from the repo docs: `RemoteSigned` at `LocalMachine` scope.)
98
99 ### 5) Create SIEM routing for action output (Graylog)
100
101 Recommended pattern:
102 - Create a dedicated **Graylog index set** (e.g., `copilot_action`)
103 - Create a **Graylog stream** that routes CoPilot Action output to that index
104
105 Why: action output is operational/response telemetry—keep it searchable without polluting core security logs.
106
107 ### 6) Add/verify Wazuh rule support for action output
108
109 Action output is typically written to active response logs that the Wazuh agent already ships.
110
111 Add/verify the supporting detection rules so the Wazuh manager can identify/classify “CoPilot Action” results.
112
113 ### 7) Import Grafana dashboards (optional but recommended)
114
115 The CoPilot-Action repo includes dashboards under:
116 - `Grafana/`
117
118 Import them into Grafana and point the datasource at the CoPilot Action index.
119
120 Why: dashboards make it much easier to confirm actions are firing and to review results across time.
121
122 ---
123
124 ## Repo pointers
125
126 Everything you need to stand this up (artifacts + dashboards + docs) lives in:
127 - https://github.com/socfortress/CoPilot-Action
128
129 ---
130
131 ## Common tasks
132
133 ### Find an action
134
135 Use search to filter actions by name/technology.
136
137 ### Invoke an action
138
139 ![Invoke action (placeholder)](../../assets/ui/agents-copilot-actions-invoke.png)
140
141 Typical flow:
142 1) Select an action
143 2) Review metadata (OS support, parameters)
144 3) Select a target agent
145 4) Invoke
146
147 Examples shown in the video:
148 - collecting browser history
149 - blocking an IP address via Windows Firewall
150
151 ### View results
152
153 Results can be viewed:
154 - in Grafana dashboards (recommended for trending/overview)
155 - in Velociraptor execution logs/results (best for deep troubleshooting)
156
157 ![Results (placeholder)](../../assets/ui/agents-copilot-actions-results.png)
158
159 ---
160
161 ## Troubleshooting (fast checks)
162
163 If an action fails:
164 - Confirm the required Velociraptor artifacts exist and are runnable
165 - Check the Velociraptor artifact execution **logs/results** for errors
166 - Verify Graylog stream/index routing if you expect results in dashboards
167 - Confirm endpoint prerequisites (PowerShell availability, script paths, permissions)
168
169 ---
170
171 ## Gotchas
172
173 - Actions can be disruptive (containment/firewall changes). Use approvals + logging.
174 - Start by testing actions on a lab endpoint and then roll out.
175 - Ensure your routing keeps action output separated from core security telemetry.
176
177 ---
178
179 ## Video context
180
181 Feature walkthrough + setup approach:
182 - https://www.youtube.com/watch?v=l9OLtgemYOQ