CTF Organizer Guide¶
How to run a capture-the-flag event end to end. The organizer surface lives under the
CTF Admin area of the portal (/ctf/admin/) and is restricted to users with CTF
permission for the requested event actions. Account membership and resource
ownership alone do not grant those permissions.
Staff changes and recovery¶
Use Event staff to assign active human or service principals as moderators, judges or co-organizers. A permitted administrator can transfer ownership with Make owner. Every change checks the current administrator's permissions.
If a change is awaiting confirmation, select Retry pending change to finish the recorded update. Check status refreshes its progress. Finish that change before starting another staff update. A current authorized administrator can resume it if the original caller loses access or the recipient is deactivated. New assignments still require active recipients.
Event Lifecycle¶
stateDiagram-v2
[*] --> Draft: Create event
Draft --> Registration: Open registration
Registration --> Live: Event start
Live --> Ended: Event end
Ended --> [*]: Cleanup
1. Create the Event¶
From CTF Admin → Events → Create, set the core parameters:
| Setting | Purpose |
|---|---|
| Name / description | Shown to participants |
| Scenario | The range scenario each participant gets |
| Event start / end | Competition window |
| Registration deadline | Latest time a participant may register |
| Max participants | Capacity limit |
| Team mode / team size | Individual vs. team competition. Team size is 1 or more; 0 or blank is unlimited |
| Range spin-up minutes | Lead time to provision ranges before start |
| Submission cooldown / attempt limit | Anti-brute-force throttles |
| Scoreboard visibility / freeze | Whether and when standings are shown |
| Live notifications | Which live notices participants get; see Notifications |
| Auto cleanup / cleanup delay | Whether ranges are torn down after the event |
| Publish public registration page | Opt-in public event details and request form; off by default |
For workspace-scoped events, the workspace supplies the egress policy for new participant and recovery-spare ranges. Each range belongs to its participant principal, or to the event owner's principal while it is a spare. Selecting a shared event workspace does not give other workspace members access to those ranges. Set the workspace policy before provisioning: changes are audited and apply to later launches, not ranges already running. If the policy changes after a spare is prepared, an incompatible spare cannot replace a participant range. For large GCP events, include both participants and spares in the shared-VPC NAT capacity check.
Optional public registration¶
Enable Publish public registration page only when you want an unauthenticated signup surface for this event. The form previews the disclosure: event name, description, start/end times, effective registration deadline, and timezone. It does not publish the scenario, workspace, capacity, participant roster, rules, challenges, briefing, custom pages, or logo. The privacy notice linked from the page is operator-supplied; review it for your deployment before sharing the URL.
The share link appears on the event overview after publication is enabled. The page is reachable only while the event is in Registration and the event remains explicitly published. Turning the switch off, starting, or cancelling the event closes the public route without changing the stored switch.
2. Add Challenges¶
Under an event, Challenges → Create. For each challenge set its name, category,
description, points, difficulty, and the flag. A static flag is stored as plaintext
without its wrapper, and participants may submit it as FLAG{value}, {value}, or
just value (the FLAG prefix is case-insensitive).
Additional controls:
- Release time: schedule a challenge to appear partway through the event.
- Prerequisite: lock a challenge until another is solved.
- Visibility: hide a challenge from the participant list until you release it.
- Target instance / port: point participants at the right host in their range.
- Files: upload challenge attachments.
- Hints: optional, point-reducing hints.
A challenge can carry multiple flags (for multi-stage solves), each validated independently.
Scenario-provided challenges¶
Some deployments bind a scenario to an immutable native CTF content bundle. Selecting that scenario while creating an event automatically populates its challenges, flags, hints, and prerequisite graph. Do not manually import or duplicate challenges for such an event.
Managed content is fail closed. An organizer edit marks the event content as drifted, and the event cannot be activated from registration. Create a new event to adopt a revised bundle instead of overwriting or merging the existing graph. The selected scenario also cannot be changed after managed content has been created.
Images in event content¶
The event logo and images in Markdown (rules, challenge descriptions and solutions, event pages, and announcements) can point to another website. The platform copies each external image into its own storage after you save the content. Participants then load the copy from the portal. The CTF scheduler copies up to five images every 30 seconds.
An image shows as a link to its original address until its copy is ready. It stays a link when the copy fails. A copy fails when the image:
- is not HTTPS
- redirects to another address
- is larger than 5 MiB
- is not a PNG, JPEG, GIF, or WebP image
The organizer image list (GET /api/v1/ctf/events/<event_id>/content-images/)
shows the state and failure reason of each image. To fix a failed image, change
the content to a direct HTTPS link to a supported image. Participants see an
image only while they can see the content that contains it.
3. Participant Briefing¶
Give participants event-specific orientation: what environment they are in, how their range is reached, and where to start. This is the guidance a shared help page cannot carry—for example, that they are a red-team operator on a Kali workstation inside a target network, reached from that box, with the public website, DNS server, and domain controller as useful first objectives.
Author it from the event's admin page in the Participant briefing editor. It is
Markdown: headings, lists, code, and links render; raw HTML, scripts, and images do
not, and unsafe link schemes (such as javascript: and data:) are stripped when the
briefing is shown. Editing updates the published briefing in place; removing it
restores the generic help.
The briefing is stored and shown verbatim in the language you write it—the platform does not translate it, so you own its wording. Because it is visible to every participant, never put flags, passwords, invitation or reset links, range credentials, or any other secret in it.
When a briefing exists, participants get a Briefing entry on their event home and a dedicated briefing page. An event with no briefing shows the standard help unchanged, so nothing regresses if you do not author one.
4. Manage Participants¶
From an event's Participants page you can add participants individually or bulk import a roster. Each participant is tracked through registration, range assignment, and scoring. Use Brackets to group participants into ranked cohorts, and (in team mode) manage team membership.
To reassign a captain, open an existing team member’s participant detail page. Select Make team captain. This action requires permission to manage teams. It changes captaincy without changing membership or scores.
Requests from the optional public page appear separately at the top of this screen. Reject closes a request without side effects. Approve sends the request through the same participant-admission path as a manual add: current capacity and duplicate-email policy are rechecked, then the isolated account, participant seat, and normal provisioning wake-up are created. A public submission itself never sends mail, creates credentials, reserves capacity, or provisions a range.
Every new participant with a delivery email gets a registration confirmation by email. This includes approved requests, single adds, CSV imports and generated accounts. The confirmation gives the event name, start, end and timezone. It never contains a password. The platform sends it only when the deployment has a mail provider. Requests can be reviewed only once and their PII is removed by the scheduler after the event-relative participant-account retention window.
Participants use isolated temporary accounts on the dedicated CTF login page. Generate accounts before an event and hand out each generated username and initial password manually, or attach an optional delivery email for non-secret login information. By default, every new account receives its own strong generated password. If an event explicitly needs one shared initial password, set it on the event form; leaving that field blank while editing preserves the existing policy, and Disable the event shared participant password returns new accounts to unique generated passwords.
Use Manage password on the participant list or detail page to generate a new
password or set a policy-compliant one. The result is displayed once and cannot be
retrieved later. Resetting a participant does not change the event-wide policy for
new accounts. A reset invalidates the participant's existing web sessions and API
tokens and requires a password change on next login. Invitation resend sends only
the login URL and username; it never resets or emails a password. Generated
usernames remain globally unique and start with range-.
Accounts stop authenticating at event end and are anonymized after the configured participant-account retention window.
5. Ranges¶
The Ranges page shows the provisioning state of every participant's range. The platform provisions ranges automatically around the spin-up window and, when auto cleanup is enabled, tears them down after the event. The CTF scheduler drives batch provisioning; see the technical CTF documentation for how it runs.
Recovering a destroyed range¶
Live-fire scenarios can leave a participant's range unrecoverable in place (the participant or their agent may encrypt, wipe, or corrupt it). From the Ranges page you can recover a single participant's range without editing any records by hand:
- Rebuild provisions a fresh range for the same event and scenario.
- Reassign spare hands the participant a prewarmed spare range from the event's spare pool, when one is available.
The old range is always destroyed so it can no longer receive the participant's traffic, and the participant keeps their scoreboard identity, submissions, awards, team and bracket membership, and registration. Recovery is safe to retry after a partial failure. The current recovery phase and any failure reason are shown on the Ranges page. A participant cannot recover their own range; recovery is organizer-only.
Setting the spare pool¶
The Spare pool panel on the Ranges page controls how many prewarmed ranges are kept ready for Reassign spare recovery. Set the target pool size and update it; the platform provisions any shortfall under managed system accounts (never counted as participants) and shows the current target, available, provisioning, and failed counts. The pool only grows on request; it does not shrink automatically. Unconsumed spares are torn down along with the rest of the event's ranges during event cleanup.
Recovery spares vs. the initial-launch warm pool¶
The event recovery-spare pool described here is distinct from the
deployment-level warm pool for faster initial launches (see
Ranges). Recovery spares replace a participant range that fails
mid-event via Reassign spare; the warm pool speeds up the first launch by
handing out a pre-provisioned range. They are configured separately (recovery
spares per event here; the warm pool in bigrae.yaml), accounted separately
against capacity, and never share generations; a recovery spare is never claimed
as an initial-launch warm range, and vice versa.
6. Notifications¶
Use Notifications to send announcements during the event.
The event sends reminder emails before the start, at the hour offsets in Reminder hours before start. New events use 24 and 1. Clear the field to turn reminders off. A change replaces the reminders that are not sent yet. Each email shows the start time in the participant's time zone. When a participant has not set a time zone, the email uses the event time zone. Reminder emails need a configured mail provider.
Live notifications in the event form select the live notices that participants get while the event runs:
| Kind | What the participant gets |
|---|---|
| Live message alerts | A toast when a new message arrives in the inbox |
| New challenge released | An inbox notice when a hidden challenge becomes visible |
| First blood | An inbox notice when a challenge has its first solve |
| Scoreboard position changes | A toast when the participant's scoreboard position changes |
| Range ready | A toast when the participant's range is ready |
A new event selects all kinds. Clear all kinds to turn live notifications off for the event. Notices do not name a challenge or a participant. First blood and position notices stop while the scoreboard is hidden or frozen.
7. During and After the Event¶
- The Scoreboard on the Monitoring page refreshes every 15 seconds. Freeze it near the end to keep final standings suspended until close.
- Analytics summarizes solves, attempts, and challenge difficulty in practice.
- After the event, scoring can be recomputed authoritatively if needed (see the technical docs for the leaderboard recompute command).
See Also¶
- CTF: the participant experience.
- CTF technical documentation: models, services, scheduling, and range provisioning.
- Native CTF scenario content: publish and bind private, digest-pinned event content.