backup: encrypted daily backups — VPS generates, Mac pulls (Tanda B, point 1)

deploy/backup/make-backup.sh (installed as /usr/local/bin/cayce-make-backup):
- pg_dump custom-format of the cayce-ops DB (URL from env or .env,
  SQLAlchemy driver qualifier stripped, credentials never logged)
- gitea dump (repos + sqlite + attachments), live; consistency notes in RESTORE.md
- config tar: Caddyfile(+d), gitea app.ini + internal_token, cayce-* units, .env
- EACH artifact encrypted with age BEFORE leaving the tmpdir; recipient
  read from /etc/cayce-backup/age-recipient.pub (public key only). The
  private age key never touches this server.
- plaintext exists ONLY in a mktemp dir removed by an EXIT trap; a partial
  day directory is removed on failure; any step failure => exit != 0 =>
  journald entry
- manifest.sha256 over the encrypted artifacts (pull side verifies
  integrity WITHOUT the age private key)
- destination /var/backups/cayce/YYYY-MM-DD/ (750 root:backup-pull, 640)
- retention: 7 daily dirs, pruned by mtime

systemd: cayce-backup.service (oneshot, hardened: ProtectSystem=strict,
NoNewPrivileges, ProtectHome=read-only) + cayce-backup.timer (daily
03:15 UTC, Persistent). Timer ENABLED on the VPS per operator instruction;
runs fail loudly (exit 1, clear journald message) until the age recipient
is provided — that is the designed not-configured state.

README.md

@@ -1,10 +1,13 @@
 # Cayce Ops
 
-Cayce Ops is the autonomous operations stack of **Cayce** — an autonomous entity
+CayceOps es el stack de operaciones de Cayce, un experimento de operaciones
-operating under the Próspera ZEDE agentic-companies jurisdiction pilot. Cayce runs
+agénticas de Cayce Co. Toda acción con efectos externos requiere aprobación
-[bitcoinhotels.org](https://bitcoinhotels.org) (Bitcoin-native travel guides) and
+humana firmada criptográficamente, con una clave que nunca reside en la
-[hardcap.news](https://hardcap.news) (Bitcoin security intelligence). The
+infraestructura de Cayce.
-experiment is managed by Cayce itself.
+
+Cayce operates [bitcoinhotels.org](https://bitcoinhotels.org)
+(Bitcoin-native travel guides) and [hardcap.news](https://hardcap.news)
+(Bitcoin security intelligence).
 
 Five decoupled components share one Postgres database. Each is a CLI command,
 driven by cron/systemd timers. Nothing runs without an audit trail.

cayceops/components/maild/classify.py

@@ -19,10 +19,17 @@ from cayceops.models import MailClassification
 
 _APPROVAL_WORDS = re.compile(r"\b(approv|accept|welcome)", re.I)
 
-# any of these tokens anywhere in the mail vetoes an `approval` result
+# any of these tokens anywhere in the mail vetoes an `approval` result.
+# English and Spanish (merchants write in both; mixed mails are common).
 _NEGATIONS = re.compile(
     r"\b(not|no|never|cannot|can'?t|won'?t|hasn'?t|isn'?t|"
-    r"no\s+longer|unable|decline\w*|regret\w*|unfortunately)\b",
+    r"no\s+longer|unable|decline\w*|regret\w*|unfortunately|"
+    # spanish: standalone negation words
+    r"nunca|jam[aá]s|tampoco|ning[uú]n(?:a|o)?|nadie|nada|"
+    # spanish: multi-word negations and rejection stems
+    r"no\s+podemos|no\s+es\s+posible|no\s+acept(?:amos|ar[aá])|"
+    r"rechaz\w*|lamentablemente|descartad\w*|"
+    r"no\s+ha\s+sido\s+aprobad\w*|no\s+procede)\b",
     re.I,
 )
 
@@ -34,11 +41,24 @@ _RULES: list[tuple[MailClassification, re.Pattern[str], float]] = [
         re.compile(r"\b(declin|reject|not able to|unable to|we regret|won.t be able)\w*", re.I),  # noqa: E501
         0.9,
     ),
+    (
+        MailClassification.REJECTION,
+        # spanish rejections (strong stems, not just negation words)
+        re.compile(r"\b(rechaz\w*|no\s+vamos\s+a\s+pod(?:er|emos)|su\s+solicitud\s+ha\s+sido\s+denegada|no\s+cumple\s+los\s+requisitos)\b", re.I),  # noqa: E501
+        0.9,
+    ),
     (
         MailClassification.APPROVAL,
         re.compile(r"\b(approv|accept|welcome to the|you.re (in|approved)|live|activated|glad to)\w*", re.I),  # noqa: E501
         0.85,
     ),
+    (
+        MailClassification.APPROVAL,
+        # spanish approvals (standalone positive stems; the veto still
+        # downgrades if ANY negation appears elsewhere in the mail)
+        re.compile(r"\b(aprob\w*|aceptad\w*|bienvenid\w*|ha\s+sido\s+aceptada|encantados?\s+de\s+(trabajar|colaborar))\b", re.I),  # noqa: E501
+        0.85,
+    ),
     (
         MailClassification.QUESTION,
         re.compile(r"\b(could you|can you|please (provide|clarify|confirm)|what (is|are)|how (do|does|can)|need more (info|details))\b", re.I),  # noqa: E501

deploy/backup/make-backup.sh

@@ -0,0 +1,119 @@
+#!/usr/bin/env bash
+# cayce-backup — encrypted daily backup of CayceOps + Gitea + config.
+#
+# Design contract (Tanda B):
+#   - EVERY artifact is encrypted with age BEFORE it leaves the tmpdir.
+#     No plaintext is ever written outside the tmpdir, which is removed
+#     by a trap even on failure.
+#   - The age recipient lives in /etc/cayce-backup/age-recipient.pub
+#     (public key only). The PRIVATE key never touches this server:
+#     decryption happens on the operator's machine after a pull.
+#   - Any step failure => non-zero exit => journald entry (run via systemd).
+#   - Retention: 7 daily directories under /var/backups/cayce/.
+#
+# The restore procedure lives in deploy/backup/RESTORE.md.
+
+set -euo pipefail
+
+DATE="$(date -u +%F)"
+BACKUP_ROOT="/var/backups/cayce"
+DEST="$BACKUP_ROOT/$DATE"
+RECIPIENT_FILE="/etc/cayce-backup/age-recipient.pub"
+CAYCE_DIR="/home/ubuntu/cayce-ops"
+ENV_FILE="$CAYCE_DIR/.env"
+RETENTION_DAYS=7
+
+log() { echo "[cayce-backup] $(date -u +%FT%TZ) $*"; }
+die() { echo "[cayce-backup] FATAL: $*" >&2; exit 1; }
+
+# --- preconditions -------------------------------------------------------------
+[[ $EUID -eq 0 ]] || die "must run as root (systemd)"
+command -v age >/dev/null || die "age not installed"
+[[ -r "$RECIPIENT_FILE" ]] || die "age recipient not configured: create $RECIPIENT_FILE (public key only)"
+RECIPIENT="$(tr -d "[:space:]" < "$RECIPIENT_FILE")"
+[[ "$RECIPIENT" == age1* ]] || die "$RECIPIENT_FILE does not look like an age1... public key"
+
+# --- workspace: everything plaintext lives here and dies here ------------------
+TMP="$(mktemp -d /tmp/cayce-backup.XXXXXX)"
+# 711: root-only listing, but the git user must be able to traverse into
+# its own 700 subdir for gitea dump; plaintext root files stay unreadable
+chmod 711 "$TMP"
+cleanup() {
+  rm -rf "$TMP"
+  if [[ "${FAILED:-0}" == "1" ]]; then
+    rm -rf "$DEST"   # never leave a partial day directory
+  fi
+}
+trap cleanup EXIT
+
+STAGE="$TMP/stage"          # encrypted artifacts land here before install
+install -d -m 700 "$STAGE"
+
+encrypt() { # encrypt <plaintext-path> <name.age>
+  age -r "$RECIPIENT" -o "$STAGE/$2" "$1"
+}
+
+# --- 1. Postgres dump (cayce-ops DB) --------------------------------------------
+DB_URL="${CAYCE_OPS_DATABASE_URL:-}"
+if [[ -z "$DB_URL" && -r "$ENV_FILE" ]]; then
+  # shellcheck disable=SC1090
+  DB_URL="$(grep -E "^CAYCE_OPS_DATABASE_URL=" "$ENV_FILE" | head -1 | cut -d= -f2-)"
+fi
+[[ -n "$DB_URL" ]] || die "no CAYCE_OPS_DATABASE_URL in env or $ENV_FILE"
+PG_URL="$(printf "%s" "$DB_URL" | sed -E "s#postgresql\+[a-z]+://#postgresql://#")"
+log "dumping database from ${PG_URL%%@*@}***@..."
+pg_dump "$PG_URL" --format=custom --file "$TMP/cayce-ops-db.dump"
+encrypt "$TMP/cayce-ops-db.dump" "cayce-ops-db.dump.age"
+log "database dump: $(du -h "$TMP/cayce-ops-db.dump" | cut -f1) plaintext, encrypted ok"
+
+# --- 2. Gitea dump (repos + sqlite + attachments) -------------------------------
+GITEA_BIN="$(command -v gitea || echo /usr/local/bin/gitea)"
+[[ -x "$GITEA_BIN" ]] || die "gitea binary not found"
+install -d -o git -g git -m 700 "$TMP/gitea-work"
+log "running gitea dump (live; see RESTORE.md for consistency notes)..."
+sudo -u git "$GITEA_BIN" dump -c /etc/gitea/app.ini \
+  --file "$TMP/gitea-work/gitea-dump.zip" --tempdir "$TMP/gitea-work" --quiet
+encrypt "$TMP/gitea-work/gitea-dump.zip" "gitea-dump.zip.age"
+log "gitea dump: $(du -h "$TMP/gitea-work/gitea-dump.zip" | cut -f1) plaintext, encrypted ok"
+
+# --- 3. Config tar (Caddy, Gitea app.ini + internal_token, units, .env) ----------
+CONF="$TMP/config"
+install -d -m 700 "$CONF"
+mkdir -p "$CONF/caddy" "$CONF/gitea" "$CONF/systemd" "$CONF/cayce-ops"
+cp -a /etc/caddy/Caddyfile "$CONF/caddy/" 2>/dev/null || true
+cp -a /etc/caddy/Caddyfile.d "$CONF/caddy/" 2>/dev/null || true
+cp -a /etc/gitea/app.ini "$CONF/gitea/" 2>/dev/null || true
+cp -a /etc/gitea/internal_token "$CONF/gitea/" 2>/dev/null || true
+find /etc/systemd/system -maxdepth 1 -name "cayce-*" \
+  -exec cp -a {} "$CONF/systemd/" \; 2>/dev/null || true
+[[ -r "$ENV_FILE" ]] && cp -a "$ENV_FILE" "$CONF/cayce-ops/dot-env" \
+  || log "WARNING: $ENV_FILE not present — config backup continues without it"
+# at least Caddy or Gitea config must exist, else the tar is useless
+[[ -e "$CONF/caddy/Caddyfile" || -e "$CONF/gitea/app.ini" ]] \
+  || die "no config sources found (Caddyfile/app.ini missing)"
+tar -C "$CONF" -czf "$TMP/config.tar.gz" .
+encrypt "$TMP/config.tar.gz" "config.tar.gz.age"
+log "config tar: $(du -h "$TMP/config.tar.gz" | cut -f1) plaintext, encrypted ok"
+
+# --- 4. install + manifest -------------------------------------------------------
+install -d -m 750 -o root -g backup-pull "$BACKUP_ROOT" 2>/dev/null \
+  || install -d -m 700 "$BACKUP_ROOT"   # backup-pull group may not exist yet
+install -d -m 750 -o root -g backup-pull "$DEST" 2>/dev/null \
+  || install -d -m 700 "$DEST"
+install -m 640 -o root -g backup-pull "$STAGE"/*.age "$DEST/" 2>/dev/null \
+  || install -m 600 "$STAGE"/*.age "$DEST/"
+( cd "$DEST" && sha256sum ./*.age > "$TMP/manifest.sha256" )
+install -m 640 -o root -g backup-pull "$TMP/manifest.sha256" "$DEST/manifest.sha256" 2>/dev/null \
+  || install -m 600 "$TMP/manifest.sha256" "$DEST/manifest.sha256"
+log "manifest written: $(wc -l < "$DEST/manifest.sha256") artifacts"
+
+# --- 5. retention -----------------------------------------------------------------
+PRUNED=0
+while IFS= read -r old; do
+  rm -rf "$old"
+  PRUNED=$((PRUNED + 1))
+  log "retention: removed $old"
+done < <(find "$BACKUP_ROOT" -mindepth 1 -maxdepth 1 -type d -name "2*-*-*" -mtime +"$RETENTION_DAYS")
+log "retention: kept last $RETENTION_DAYS days (pruned $PRUNED)"
+
+log "BACKUP OK → $DEST ($(du -sh "$DEST" | cut -f1))"

deploy/systemd/cayce-backup.service

@@ -0,0 +1,14 @@
+[Unit]
+Description=Cayce — encrypted daily backup (CayceOps DB + Gitea + config)
+After=network-online.target postgresql.service
+Wants=network-online.target
+
+[Service]
+Type=oneshot
+ExecStart=/usr/local/bin/cayce-make-backup
+# hardening
+NoNewPrivileges=true
+ProtectHome=read-only
+ReadWritePaths=/var/backups/cayce /tmp
+ProtectSystem=strict
+PrivateTmp=false

deploy/systemd/cayce-backup.timer

@@ -0,0 +1,10 @@
+[Unit]
+Description=Cayce — daily backup schedule (03:15 UTC)
+
+[Timer]
+OnCalendar=*-*-* 03:15:00 UTC
+Persistent=true
+RandomizedDelaySec=300
+
+[Install]
+WantedBy=timers.target

tests/test_spanish_negations.py

@@ -0,0 +1,80 @@
+"""Spanish negation tests (round-3 point 4): merchants write in Spanish;
+negations like 'nunca', 'lamentablemente' or 'no podemos' must veto
+`approval` exactly like their English counterparts. Veto rule unchanged:
+when in doubt -> needs_review."""
+
+from __future__ import annotations
+
+import pytest
+
+from cayceops.components.maild import classify as clf
+from cayceops.models import MailClassification
+
+APPROVAL_ONLY = MailClassification.APPROVAL
+
+
+def _never_approval(cfg_factory, text: str) -> None:
+    kind, _ = clf.classify(text, cfg_factory())
+    assert kind is not APPROVAL_ONLY, f"{text!r} classified as APPROVAL"
+
+
+def test_english_baseline_still_works(cfg):
+    config = cfg()
+    kind, _ = clf.classify("Your application was approved, welcome to the program", config)
+    assert kind is APPROVAL_ONLY
+
+
+@pytest.mark.parametrize(
+    "text",
+    [
+        # 1. rejection stem
+        "Lamentamos informarle que su solicitud fue rechazada",
+        # 2. nunca
+        "Nunca aceptamos negocios de criptomonedas",
+        # 3. jamás
+        "Jamás hemos trabajado con affiliates externos",
+        # 4. tampoco
+        "No podemos aceptar su sitio, tampoco podemos ofrecerle otro acuerdo",
+        # 5. ningún/ninguna
+        "En este momento no tenemos ningún hueco para nuevos partners",
+        # 6. no podemos (multi-word)
+        "Su propuesta es interesante pero no podemos proceder",
+        # 7. no es posible (multi-word)
+        "Gracias por escribir, pero no es posible colaborar este trimestre",
+        # 8. lamentablemente
+        "Lamentablemente su aplicación no ha sido aprobada",
+        # 9. no aceptamos
+        "Por política interna no aceptamos sitios de viajes",
+        # 10. mixed spanish/english: negation in spanish, approval word in english
+        "Thanks for reaching out! Lamentablemente we cannot approve your site",
+        # 11. mixed: approval in spanish, negation in english
+        "Apreciamos su interés, but we are not accepting new partners",
+        # 12. nadie / nada
+        "Aquí nadie decidiría eso, así que no hay nada que aprobar",
+    ],
+)
+def test_spanish_and_mixed_negations_veto_approval(cfg, text):
+    _never_approval(cfg, text)
+
+
+def test_spanish_pure_approval_still_approves(cfg):
+    """A clean Spanish approval WITHOUT any negation must still classify
+    approval — the veto must not over-fire."""
+    config = cfg()
+    kind, confidence = clf.classify(
+        "Su solicitud fue aprobada. Bienvenido al programa, estamos encantados "
+        "de trabajar con ustedes.",
+        config,
+    )
+    assert kind is APPROVAL_ONLY
+    assert confidence >= 0.5
+
+
+def test_spanish_rejection_classifies_rejection(cfg):
+    config = cfg()
+    kind, conf = clf.classify(
+        "Lamentamos informarle que su solicitud fue rechazada por el comité",
+        config,
+    )
+    assert kind is MailClassification.REJECTION
+    assert conf >= 0.9