Skip to documentation

Security and privacy

Follow practical operator safeguards for least privilege, sensitive evidence, consent, exports, and safe support requests.

Audience
moderator · senior reviewer · team lead · server admin · public appellant
Availability
free
Maturity
stable

Last reviewed

On this page

Everyday operator safeguards

  • Confirm the selected server, target, channel, and record before every write.
  • Use the least-privileged Discord role and ModFlow team that can complete the job.
  • Keep evidence, staff notes, transcripts, and exports inside their guarded workflow.
  • Treat copied links and screenshots as sensitive; they never grant permission.
  • Review consent and artifact status before using recorded-session material.
  • Sign out of shared devices and never share credentials, tokens, sessions, or private keys.

When something sensitive is exposed

  1. Stop sharing

    Remove the material from further messages or exports where you are authorized to do so.

    Check: Exposure is not widened.

  2. Preserve a safe record

    Record server, time, affected surface, and non-sensitive identifiers without copying the content.

    Check: The issue can be investigated without a second disclosure.

  3. Notify the right server lead

    Use the server's established staff escalation path for moderation data or permission problems.

    Check: Server-scoped containment begins.

  4. Use the published contact

    Report a product security/privacy concern with the minimum necessary detail.

    Check: The concern reaches the documented contact path.

Recognize common operator risks

Common risks
RiskExampleControl
Wrong server/targetActing on a similarly named member or server.Verify numeric IDs before every write.
Over-privileged roleGiving broad Administrator access to fix one denied page.Grant the narrow team/Discord capability and test denied neighbors.
Evidence leakagePasting senior-only evidence into a general staff note.Share the guarded Case reference and safe summary.
Session compromiseShared browser, copied cookie, forwarded OAuth link.Use individual sign-in, lock devices, sign out of shared systems.
Export sprawlUnmanaged CSV/JSON copies in personal storage.Purpose limitation, secure destination, controlled deletion.
Consent failureRecording a participant without verified implemented consent.Use Soundcheck notice/consent and stop if state is unclear.
Duplicate punishmentRetrying after an interaction timeout without checking records.Verify Discord state and stored execution first.
Support oversharingSending tokens, transcript, or evidence file with a bug report.Send safe identifiers/status only.

Use a shift security checklist

  1. Start secure

    Confirm personal staff account, controlled device, intended server, and current role.

    Check: The session and scope are trusted.

  2. Minimize access

    Open only records and downloads needed for the active job.

    Check: Sensitive exposure is reduced.

  3. Verify before write

    Check target/server/action/options and current state.

    Check: Wrong or duplicate actions are prevented.

  4. Keep artifacts guarded

    Use Case, Voice Evidence, Notes, and Export boundaries.

    Check: Private content remains in controlled surfaces.

  5. Handoff minimally

    Share record IDs, safe status, and next owner.

    Check: Chat does not become a second evidence store.

  6. End secure

    Close evidence/downloads, remove temporary copies, sign out on shared devices.

    Check: The shift leaves no unnecessary access.

Respond to specific security/privacy incidents

Incident response first steps
IncidentFirst actionPreserve
Wrong member actionStop further actions and use supported reversal.Action/execution IDs, current Discord state, correction reason.
Evidence shown to wrong roleRemove/contain sharing and correct visibility/permission.Record ID, viewer role, time, non-sensitive exposure scope.
Export sent to wrong destinationStop sharing/delete where authorized and notify server lead.Dataset, filters, time, recipient scope—not the data itself in support.
Session/account suspected compromisedSign out/revoke through official account process and stop writes.Times, routes, safe audit identifiers.
Recording consent unclearStop recording/participation workflow.Session ID, notice/consent/status events.
Unexpected excluded feature exposedDo not experiment; use stable alternatives.Surface name, server, viewer role, status, screenshot without private data.

Prepare a secure support report

Include versus exclude
IncludeExclude
Server name and numeric IDDiscord bot token or OAuth secret
Page/command/App/reaction nameSession cookie or browser storage
Case/action/appeal/session/job IDPrivate evidence or full transcript
Exact visible status/error and timeExport file or payment details
Last safe reproduction stepUnnecessary member personal data
Redacted screenshot only when safeHidden-channel screenshots or credentials

Protect providers, destinations, and copies

  • Keep Discord log, Jail, Watchlist, Ban Review, and assignment destinations private to their intended participants.
  • Use provider authorization screens; never paste OAuth refresh tokens, API keys, confidential bearer links, or exported evidence into support messages.
  • Select the minimum Google or Microsoft file needed, and disconnect a provider when its assignment work is complete.
  • For any approved MIRA run, minimize sources, review every citation, and do not treat generated text as fact or punishment authority.
  • Track downloaded exports and rehosted artifacts because later ModFlow deletion or permission changes cannot retract external copies.
Back to top