What changed on my servers' files
Which files is Wazuh watching, and what changed?
Wazuh records every file it watches on your servers. Get the total per server, and the files that changed most recently, without logging in to any of them.
What you get: how many files Wazuh’s file integrity monitoring tracks on each server, which servers are still reporting, and the most recently changed files with their owner, permissions and size. Nothing on your servers changes.
Ask Quox
Section titled “Ask Quox”Can you show me my file integrity in Wazuh? I'd like to know the total.
▤ AOCL │ L1 identity complete · allowed · 1ms
▤ AOCL │ L3 policy gate complete · proceed · 8ms
▤ AOCL │ handoff narration kept out of the answer · I'll hand this to ARGUS, our Wazuh specialist.
◐ DELEGATE │ delegate_to_agent agent_id=argus ✔ DELEGATE │ delegate_to_agent agent_id=argus │ 3 tools used, 3 tool outcomes · 31.2s
◐ ARGUS › WAZUH │ agents ✔ ARGUS › WAZUH │ agents │ 70ms
◐ ARGUS › WAZUH │ fim ✔ ARGUS › WAZUH │ fim │ 280ms
◐ ARGUS › WAZUH │ fim ✔ ARGUS › WAZUH │ fim
⛓ AEE │ envelope opened · ops.tool · delegate_to_agent ok
⛓ AEE │ envelope opened · stream_success
I read the file integrity data from all three Wazuh agents. Wazuh holds 10,565 file integrity entries in total: 6,783 on your 2 agents and 3,782 on the manager (000) (source: wazuh_fim). Every read succeeded, so none of these counts stands in for a failed read. Agent 001 Name hyp-web-01 Connection disconnected, last check-in 2026-11-11 15:03 UTC FIM entries 3,367 Agent 002 Name hyp-db-01 Connection disconnected, last check-in 2026-11-11 15:06 UTC FIM entries 3,416 Agent 000 Name hyp-wazuh-01 (the manager) Connection active FIM entries 3,782 (connection state and last check-in from wazuh_agents; entry counts from wazuh_fim) What these numbers mean • These are files being watched, not changes. Wazuh keeps a record of each watched file: its size, permissions, owner and checksums. The 10,565 figure counts those records, so it does not tell you how many files changed. I can't give an honest count of changed files from this data alone. To get one, I'd look at the file integrity alerts (added, modified and deleted files) for a set time window. • Both of your agents are disconnected. hyp-web-01 and hyp-db-01 last checked in about 4 hours 45 minutes ago (source: wazuh_agents). Their file data was last updated around then, so any file changes after about 15:05 UTC are invisible to Wazuh. I'd treat that gap as the more urgent finding. My confidence is high, because it comes straight from the agent connection records. Next steps (all read-only) 1. Group the file integrity alerts for the last 24 hours by rule and host. That gives the real changed-file count, with rule levels and MITRE IDs. 2. Look into why agents 001 and 002 went quiet within three minutes of each other. When two hosts drop at almost the same time, the cause is usually something they share, such as the network path or the manager's listener, rather than each host failing on its own. Nothing was changed. Every read is recorded in this session's evidence trail. Tools used: wazuh_agents, wazuh_fim ×2 Wazuh holds 10,565 file integrity entries in total, and both of your agents have been disconnected since about 15:05 UTC.
39.8s · 4 tools · 0 approvals
quox.✔ Machine verified against the CLI on 11 Nov 2026 · test report mwcj7u
ARGUS, the Wazuh specialist, read the agent list and the file integrity records for all three agents: 10,565 entries in total, 3,367 on hyp-web-01, 3,416 on hyp-db-01 and 3,782 on the manager. It was careful about what that number means. These are records of watched files, not a count of changes, and it said it could not give an honest changed-file count from this data alone. It also flagged that both servers had been disconnected for nearly five hours, so file changes after about 15:05 UTC were invisible to Wazuh, and ranked that as the more urgent finding. Its next steps were both reads (group the file integrity alerts by rule and host, look into why both agents went quiet within three minutes of each other); it changed nothing.
Do it yourself
Section titled “Do it yourself”1. List the most recently changed files
Section titled “1. List the most recently changed files”quox wazuh fim --limit 5Ⓠ QUOX │ wazuh fim │ showing 10 of 7198 across 2 agents ✔WHEN UTC AGENT TYPE SIZE OWNER PERM FILE5d 17h ago 2026-11-06 02:57 UTC hyp-wazuh-01 file 12977 root:root rwxr-xr-x /usr/bin/podebconf-display-po5d 17h ago 2026-11-06 02:57 UTC hyp-wazuh-01 file 32524 root:root rwxr-xr-x /usr/bin/podebconf-report-po5d 17h ago 2026-11-06 02:57 UTC hyp-wazuh-01 file 301 root:root rwxr-xr-x /usr/sbin/update-grub-gfxpayload5d 18h ago 2026-11-06 02:52 UTC hyp-wazuh-01 file 874 root:root rw-r--r-- /etc/fonts/conf.avail/20-unhint-small-dejavu-lgc-sans-mono.conf5d 18h ago 2026-11-06 02:52 UTC hyp-wazuh-01 file 866 root:root rw-r--r-- /etc/fonts/conf.avail/20-unhint-small-dejavu-lgc-serif.conf5d 17h ago 2026-11-06 03:05 UTC hyp-db-01 file 301 root:root rwxr-xr-x /usr/sbin/update-grub-gfxpayload5d 17h ago 2026-11-06 03:05 UTC hyp-db-01 file 874 root:root rw-r--r-- /etc/fonts/conf.avail/20-unhint-small-dejavu-lgc-sans-mono.conf5d 17h ago 2026-11-06 03:05 UTC hyp-db-01 file 864 root:root rw-r--r-- /etc/fonts/conf.avail/20-unhint-small-dejavu-lgc-sans.conf5d 17h ago 2026-11-06 03:05 UTC hyp-db-01 file 866 root:root rw-r--r-- /etc/fonts/conf.avail/20-unhint-small-dejavu-lgc-serif.conf5d 17h ago 2026-11-06 03:05 UTC hyp-db-01 file 866 root:root rw-r--r-- /etc/fonts/conf.avail/20-unhint-small-dejavu-sans-mono.confShowing 10 of 7198; raise --limit to see more. The CLI reads every active agent by default. By this run hyp-db-01 had reconnected and hyp-web-01 had not, so it totals 7,198 across the manager and hyp-db-01, where the conversation counted all three agents.
2. Narrow it down
Section titled “2. Narrow it down”--agent <id|name> reads one server, --file <path> checks a single file such as /etc/passwd, and --search <text> searches the records for some text.
What just happened
Section titled “What just happened”Quox read Wazuh’s agent list and its file integrity records. Reading changes nothing, so no approval was needed. Anything that would change a server, such as an active response, waits for a person to approve it.