| name | on-leave |
| description | Assess alliance members for On Leave community status using fleet attendance, killmails, and Discord voice, then produce a reviewer report and admin upload CSV. Use when putting people on leave, reviewing inactive alliance members, stripping ping status for OPSEC or absence, auditing who is not participating in alliance fleets, or running a leave hygiene pass. |
On Leave
Imposed leave for Alliance members who are not participating in alliance
fleets. Decide leave vs keep from metrics; deliver a report + upload CSV.
Do not change status yourself — the executor uploads after review.
Also read debug-production-readonly-db.
Policy
Generally, if they're not participating in alliance fleets, we put them on leave.
- Away: takes away ping status so they're not getting constant pings and feel
the guilt of logging in.
- OPSEC: for people in alliance but not participating, removes ping status;
they have to talk to a CEO to restore it.
Tag every recommend Away or OPSEC. Same on_leave status either way.
on_leave → Django On Leave group only (affiliation role stripped) via
sync_user_community_groups. Most PilotFeatures deny on_leave.
Imposed leave only — do not invent declared LOA requests.
Quick start
cd backend
pipenv run python ../.cursor/skills/on-leave/scripts/fetch_alliance_activity.py --json --max-fleets 6
Defaults: 90d window; soft line ~1 fleet/month (≈ 3 fleets / 90d).
Override only from session context (quiet war window, user-named exempts).
Workflow
Task Progress:
- [ ] Fetch (--max-fleets 6)
- [ ] Drop auto-exempts (staff, directors) + rejoin/restore grace + named exempts
- [ ] Decide leave/keep; Story + Conf + reason
- [ ] Emit report + CSV
- [ ] Stop for executor review/upload
Scope
| Filter | Value |
|---|
| Affiliation | Alliance (check prod name if fetch is empty) |
| Status | active only (previous_status from fetch) |
| Activity | All linked EveCharacters (account-level) |
Auto-exempt (omit from recommend):
- Auth groups:
People Team, Technology Team, Tribe - Chief
- Corporation directors: any
Corp … Director / Corporation Director auth
group, or user on EveCorporation.directors
- Named CEOs / LTI this session
Recent rejoin grace (omit from recommend): first fleet in the 90d window
falls within the last 30 days, and no fleets in the 30–180d band before
that (returning after a long gap). Example: one fleet yesterday after months
dark → keep, do not flip.
Recent restore grace (omit from recommend): any
UserCommunityStatusHistory row in the last 30 days with
from_status=on_leave → to_status=active (or other non-leave). Someone a CEO
just took off leave should not be flipped back on the next hygiene pass. Fetch
exposes restored_from_leave_at (ISO date or null); treat non-null as omit.
Signals
| Signal | Model | Roll-up |
|---|
| Fleets | EveFleetInstanceMember | Distinct instances across linked EVE IDs |
| Kills | EveCharacterKillmailAttacker | Attackers; exclude self-victim |
| Voice | DiscordChannelActivityRecord | voice_minute by Django username → hours |
| Restore | UserCommunityStatusHistory | Latest on_leave→non-leave in last 30d → restored_from_leave_at |
Fleets are primary. Kills/voice set Away vs OPSEC and break ties. Login,
mining, PI, industry do not clear the fleet bar unless this run expands
participation. Note untracked fleets / ESI gaps / zero linked chars in reasons
when they matter.
Decision rules (90d)
- Exempt / director / recent-rejoin grace / recent-restore grace → keep (omit).
- Fleets ≥ 6 → keep (even low kills).
- Fleets 0–2, weak support (under ~5 kills and under ~5h voice) → recommend.
Away if kills≈0 and voice≈0; else OPSEC. Conf high.
- Fleets 0–2, strong kills (~15+) or voice (~10h+) → recommend OPSEC.
Conf medium if killboard-heavy (untracked ops possible).
- Fleets 3–5 → keep if ~5+ kills, ~5h+ voice, or clear support pattern;
else recommend, Conf medium.
- No linked characters → recommend, Conf medium, caveat in reason.
Bias: leave for 0–2 fleets; keep for 3+. Quiet alliance (already stated) → treat
0–3 like 0–2. Resolve every case — no borderline dump, no per-pilot questions.
Reason line: Story + 1–2 metrics (+ caveat if needed). See examples.md.
Deliverables
Always both. Sort recommends by Conf (high then medium).
Summary: N recommended (H/M); K kept; E exempt. Window: 90d.
Report
| Username | Primary | Previous status | New status | Fleets (90d) | Kills (90d) | Voice | Story | Conf | Reason |
|---|
- Previous = fetch
previous_status (normally active)
- New =
on_leave
- Voice as hours (
12.5h)
- Kept/exempt: counts only unless asked
CSV (admin upload)
Recommended rows only:
username,community_status,reason
someuser,on_leave,Away — 0 fleets, 0 kills, 0h voice (90d)
otheruser,on_leave,OPSEC — 0 fleets, 2 kills, 25h voice
community_status = on_leave. reason ≤255. Username = Django username.
Previous/New are report-only.
| Flag | Use |
|---|
--days 90 | Lookback |
--max-fleets 6 | At/below active-enough line |
--affiliation Alliance | AffiliationType.name |
Executor: apply CSV
- Review report (Previous → New, metrics, Story/Conf/Reason).
- Delete CSV rows you want to keep active; save
on_leave_YYYY-MM-DD.csv (UTF-8).
- Admin:
/admin/groups/usercommunitystatus/bulk-upload/
- Upload file; optional default reason (prefer per-row reasons).
- Upload and apply → Celery background; Discord role sync may take minutes.
- Spot-check Community Status (
on_leave) and Discord.
Bulk upload only (history + group sync). No prod shell writes.
Related
