trinetra.
cyber defense — learn
tutorials · lab 09 · soc log triage
09 / 10 · INFO
04step · module overview

You finished Lab 09. Here's what stuck.

A short recap of the moves you used in the SOC pod — keep them close. Blue team is detective work, and the next lab flips your role: in Lab 10 you become the adversary, probing an LLM that's been told never to talk.

Module complete · soc-logs

You read the alert, joined four log sources on a session ID, and reconstructed a four-stage intrusion timeline. Blue team is detective work — every incident is already written down in log lines that exist; your job is to know which four to read, and in what order.

XP earned+275xp
DifficultyMedium
Time spent~45min
TrackInfo
Recap · the five beats keep close
  1. 01 The alert is the trailhead. A SIEM correlation rule fires — one yellow square on a dashboard, a ticket number, a time window. That is the start of the story, not the end. Pull the alert metadata, open a ticket, then walk the path.
  2. 02 Failed-then-successful logins are the bell. Start at auth.log. 847 failures from one IP followed by one acceptance is the textbook brute-force shape. Pin two evidence fields: brute_user and brute_src. They're the first two pieces of the report.
  3. 03 Pivot to what they did after the door opened. The successful login gave you a session ID. Join it against access.log and you see the post-compromise path: admin dashboard, user list, sensitive file read. Pin the third evidence field: read_path.
  4. 04 Outbound bytes to nowhere should always ring the alarm. Check firewall.log for large transfers to non-RFC1918 destinations near the incident window. 11MB to a TOR exit, matching the export file size — that's the fourth evidence field: exfil_dst.
  5. 05 Four log lines, one story — submit the flag. User · source IP · file path · exfil destination, joined with underscores: TrinetraCTF{...}. The flag IS the incident report. If you can assemble it, you can write the post-mortem and brief the IR lead.
Next module

Lab 10 · AI Prompt Injection

Attack a FastAPI chatbot that's been told never to reveal a secret. Probe for OWASP LLM01.

start · lab 10 →