fix: preserve run provider capability

hub/deploy/NEW_SILO_RUNBOOK.md

@@ -0,0 +1,131 @@
+# New Alpha Silo runbook
+
+Use this runbook for a new Organization. A Silo is not merely another row in
+the existing database: it has an independent PostgreSQL role/database, Linux
+service identity, systemd unit, secret directory/keyring, workspace, skill
+store, loopback port, domain, Feishu app and provider credential.
+
+The repeatable entry point is:
+
+```sh
+bash hub/deploy/new_silo.sh
+```
+
+After collecting the Organization inputs, the wizard shows the assigned
+resources and asks once before applying them directly over SSH. The generated
+bundle is only a root-secret-safe retry and audit checkpoint; the operator does
+not execute it manually during the normal path.
+
+It gathers values and writes a private deployment bundle below
+`~/.cph-silo-plans/<instance-id>/`. The directory and all generated files are
+mode `0700`/`0600`. Never commit, paste into chat, or copy that directory into
+an immutable release. Run the wizard once per Organization; do not edit a copy
+from another Organization.
+
+## Inputs to collect
+
+The wizard derives the instance/Organization id from the slug, uses the current
+release and managed Alpha defaults, connects to the managed host (currently
+`39.107.254.4`), derives
+`https://<organization-slug>.educraft.paradigm-edu.net` from the wildcard DNS,
+and allocates an unused loopback port plus short workspace path by inspecting
+existing Silo environments, listening sockets and workspace paths over
+read-only SSH. The Organization administrator supplies:
+
+- Organization display name and slug;
+- Feishu App ID and App Secret (the wizard resolves the bot Open ID);
+- the first OWNER's Open ID and display name;
+- an Organization-exclusive OpenRouter token.
+
+Everything else is platform-managed or derived: instance/Organization id,
+server, SSH settings, release, resource ceilings, database coordinates and
+generated password, domain, port, workspace, provider/base URL, model/role,
+curated skills, concurrency, request/file limits and the managed Mihomo proxy
+environment.
+
+The Feishu app is scoped to this Silo. OAuth users authenticated by that app are
+automatically admitted to this Organization; OWNER remains the initial
+privileged membership used for controlled administration and bootstrap. An
+empty initial team list does not block the Alpha.
+
+To target a replacement platform-managed host, the platform operator may set
+`CPH_ALPHA_HOST`, `CPH_ALPHA_DEPLOY_USER`, `CPH_ALPHA_SSH_PORT` and
+`CPH_ALPHA_BASE_DOMAIN` before running the wizard. These are fleet controls,
+not Organization setup questions.
+
+Obtain a person's Open ID from the Feishu user-get documentation page by
+clicking the `user_id` value picker and selecting the person. Configure the
+redirect URL shown by the generated `OPERATE.md`; it is required for first-time
+OAuth admission.
+
+## Host prerequisites
+
+Before the first Silo on a host, install Node.js 24+, npm, rsync, PostgreSQL
+server/client, `pg_isready`, systemd, bubblewrap, socat, `runuser`, `setpriv`,
+`pg_dump`, tar, sha256sum, Nginx, Certbot, Typst and a compatible `cph` binary.
+Configure outbound proxying independently at host/service level and verify both
+GitHub and the selected model provider through it. The wizard does not install
+or select proxy nodes.
+
+Use a deployment account with only the required passwordless sudo operations.
+The application itself always runs as the installer-created non-root
+`cph-<instance-id>` user. PostgreSQL must have a separate login role and logical
+database per Silo even when all Silo databases share one PostgreSQL server.
+
+## Execute a generated bundle
+
+Open the bundle's `OPERATE.md` and perform its numbered gates in order:
+
+1. DNS, Feishu redirect URL, permissions and event subscription.
+2. Dedicated PostgreSQL role and database.
+3. Immutable release publication. Exit 78 is expected only when the first
+   installer call seeds this instance's keyring and environment template.
+4. Root-owned secret installation and off-host keyring recovery copy.
+5. Prisma migration, stopped service installation and idempotent bootstrap.
+6. Explicit runtime role/skill installation.
+7. Nginx/TLS, service start, internal/external health and Feishu acceptance.
+8. First off-host backup.
+
+Every command must fail fast. Do not add `|| true` around install, migration,
+bootstrap, Nginx validation, health, or backup checks. If a check fails, retain
+the unit logs and the exact failed stage before changing configuration.
+
+## Default runtime role and skills
+
+Roles and skills are dynamic Silo state, not release contents. The release
+contains only the management CLI. Stage approved skill directories on the host
+and install them with `agent_config.sh`; PostgreSQL records role definitions and
+skill selections while the versioned skill content lives in the Silo state
+directory and is included in backups.
+
+For the current Alpha, upsert the default role with the agreed education
+assistant system prompt, model selection, tools policy and approved skill list.
+Keep the prompt in a root-controlled staging file, pass it via
+`--system-prompt-file`, then verify with `agent_config.sh list`. Do not sync an
+operator's entire personal skill directory: each enabled skill must be reviewed
+and named explicitly. Typst being installed on the host and the Typst skill
+being enabled are separate gates.
+
+## Acceptance gate
+
+A Silo is ready only when all of the following pass:
+
+- its systemd service is active and both loopback and TLS health endpoints pass;
+- startup preflight sees exactly the configured Organization plus active Feishu
+  and provider connections;
+- OWNER completes OAuth and can interact with the bot;
+- a non-OWNER completes OAuth and can interact with the same app/Organization;
+- two Feishu groups bind distinct projects/sessions;
+- a restart preserves persisted session cursor behavior;
+- the configured provider/model succeeds through the host proxy;
+- every enabled skill is listed, and a Typst task succeeds if Typst is enabled;
+- the first business backup and separate recovery backup are stored off-host.
+
+## Rollback boundary
+
+For a failed code release, point only this instance back to the previous
+immutable release and rerun its installer with the same instance parameters.
+Do not roll back a database after migrations unless that release's documented
+database compatibility permits it. Preserve the environment, keyring, database,
+workspace and skill store. For destructive recovery, stop traffic and use the
+separate restore procedure; never substitute another Silo's state.

hub/deploy/README.md

@@ -1,5 +1,16 @@
 # Alpha Silo service installation
 
+For a brand-new Organization, start with the repeatable wizard and end-to-end
+operator runbook:
+
+```sh
+bash hub/deploy/new_silo.sh
+```
+
+See [NEW_SILO_RUNBOOK.md](NEW_SILO_RUNBOOK.md). The remainder of this document
+describes the individual installer and maintenance primitives used by the
+generated bundle.
+
 The supervised alpha runs one Organization per named Silo. The supported host
 has systemd, PostgreSQL, Node.js 24+, `pg_isready`, `runuser`, `setpriv`,
 bubblewrap, `socat`, `pg_dump`, `tar`, `sha256sum`, and a compatible `cph`.

hub/deploy/apply_new_silo.sh

@@ -0,0 +1,73 @@
+#!/usr/bin/env bash
+# Apply a bundle produced by new_silo.sh to the managed Alpha host.
+set -euo pipefail
+
+BUNDLE="${1:?usage: apply_new_silo.sh BUNDLE_DIR}"
+ANSWERS="$BUNDLE/answers.env"
+[ -f "$ANSWERS" ] || { echo "missing $ANSWERS" >&2; exit 1; }
+
+value() { sed -n "s/^$1=//p" "$ANSWERS" | tail -n1; }
+INSTANCE_ID="$(value INSTANCE_ID)"
+ORG_ID="$(value ORGANIZATION_ID)"
+HOST="$(value DEPLOY_HOST)"
+SSH_USER="$(value DEPLOY_USER)"
+SSH_PORT="$(value DEPLOY_SSH_PORT)"
+SSH_KEY="$(value DEPLOY_SSH_KEY)"
+BASE="$(value DEPLOY_BASE)"
+RELEASE="$(value RELEASE_ID)"
+HUB_PORT="$(value HUB_PORT)"
+WORKSPACE="$(value WORKSPACE_ROOT)"
+MEMORY="$(value MEMORY_MAX)"
+CPU="$(value CPU_QUOTA)"
+TASKS="$(value TASKS_MAX)"
+DB_NAME="$(value DATABASE_NAME)"
+DB_USER="$(value DATABASE_USER)"
+DB_PASSWORD="$(value DATABASE_PASSWORD)"
+PUBLIC_URL="$(value PUBLIC_BASE_URL)"
+DOMAIN="${PUBLIC_URL#https://}"
+HUB_DIR="$BASE/releases/$RELEASE/hub"
+ENV_PATH="$BASE/.secrets/$INSTANCE_ID/platform.env"
+KEYRING_PATH="$BASE/.secrets/$INSTANCE_ID/secret-keyring.json"
+UNIT="cph-hub-$INSTANCE_ID.service"
+SSH=(ssh -i "$SSH_KEY" -p "$SSH_PORT" -o BatchMode=yes "$SSH_USER@$HOST")
+SCP=(scp -i "$SSH_KEY" -P "$SSH_PORT")
+
+for file in platform.env bootstrap.json default-role-prompt.md; do
+  [ -f "$BUNDLE/$file" ] || { echo "missing bundle file: $file" >&2; exit 1; }
+done
+
+echo "[1/8] Verify immutable release"
+"${SSH[@]}" "test -f '$BASE/releases/$RELEASE/.complete'"
+
+echo "[2/8] Create dedicated database"
+if ! "${SSH[@]}" "sudo -u postgres psql -Atqc \"select 1 from pg_database where datname='$DB_NAME'\"" | grep -qx 1; then
+  printf "CREATE ROLE %s LOGIN PASSWORD '%s';\nCREATE DATABASE %s OWNER %s;\n" \
+    "$DB_USER" "$DB_PASSWORD" "$DB_NAME" "$DB_USER" | "${SSH[@]}" sudo -u postgres psql -v ON_ERROR_STOP=1
+fi
+
+echo "[3/8] Seed instance keyring and service template"
+set +e
+"${SSH[@]}" "BASE='$BASE' HUB_DIR='$HUB_DIR' INSTANCE_ID='$INSTANCE_ID' WORKSPACE_ROOT='$WORKSPACE' PORT='$HUB_PORT' MEMORY_MAX='$MEMORY' CPU_QUOTA='$CPU' TASKS_MAX='$TASKS' bash '$HUB_DIR/deploy/install_service.sh'"
+status=$?
+set -e
+[ "$status" -eq 0 ] || [ "$status" -eq 78 ] || exit "$status"
+
+echo "[4/8] Upload root-only configuration"
+remote_stage="/root/.cph-bootstrap-$INSTANCE_ID"
+"${SSH[@]}" "install -d -o root -g root -m 0700 '$remote_stage'"
+"${SCP[@]}" "$BUNDLE/platform.env" "$BUNDLE/bootstrap.json" "$BUNDLE/default-role-prompt.md" "$SSH_USER@$HOST:$remote_stage/"
+"${SSH[@]}" "install -o root -g root -m 0600 '$remote_stage/platform.env' '$ENV_PATH'; chmod 0600 '$remote_stage/bootstrap.json' '$remote_stage/default-role-prompt.md'"
+
+echo "[5/8] Migrate, install, and bootstrap"
+"${SSH[@]}" "set -euo pipefail; set -a; . '$ENV_PATH'; set +a; node '$HUB_DIR/node_modules/prisma/build/index.js' migrate deploy --schema '$HUB_DIR/prisma/schema.prisma'; BASE='$BASE' HUB_DIR='$HUB_DIR' INSTANCE_ID='$INSTANCE_ID' WORKSPACE_ROOT='$WORKSPACE' PORT='$HUB_PORT' MEMORY_MAX='$MEMORY' CPU_QUOTA='$CPU' TASKS_MAX='$TASKS' bash '$HUB_DIR/deploy/install_service.sh'; node '$HUB_DIR/dist/deployment/bootstrap-silo-cli.js' --config-file '$remote_stage/bootstrap.json' --keyring-file '$KEYRING_PATH'"
+
+echo "[6/8] Copy curated skills and configure default role"
+"${SSH[@]}" "set -euo pipefail; stage='/var/lib/cph-hub/$INSTANCE_ID/state/operator-staging'; install -d -o cph-$INSTANCE_ID -g cph-$INSTANCE_ID -m 0700 \"\$stage/skills\"; install -o cph-$INSTANCE_ID -g cph-$INSTANCE_ID -m 0600 '$remote_stage/default-role-prompt.md' \"\$stage/default-role-prompt.md\"; for source in /var/lib/cph-hub/para-26071100/state/skills/versions/*; do name=\$(sed -n 's/^name: *//p' \"\$source/SKILL.md\" | head -n1); case \"\$name\" in outline|lesson-project|data-processing-spec|typst) cp -a \"\$source\" \"\$stage/skills/\$name\"; chown -R cph-$INSTANCE_ID:cph-$INSTANCE_ID \"\$stage/skills/\$name\";; esac; done; for name in outline lesson-project data-processing-spec typst; do INSTANCE_ID='$INSTANCE_ID' ENV_FILE='$ENV_PATH' HUB_DIR='$HUB_DIR' bash '$HUB_DIR/deploy/agent_config.sh' install-skill --organization '$ORG_ID' --source \"\$stage/skills/\$name\" --version 1; done; INSTANCE_ID='$INSTANCE_ID' ENV_FILE='$ENV_PATH' HUB_DIR='$HUB_DIR' bash '$HUB_DIR/deploy/agent_config.sh' upsert-role --organization '$ORG_ID' --role draft --label '智能助手' --system-prompt-file \"\$stage/default-role-prompt.md\" --tools-json null; INSTANCE_ID='$INSTANCE_ID' ENV_FILE='$ENV_PATH' HUB_DIR='$HUB_DIR' bash '$HUB_DIR/deploy/agent_config.sh' set-role-skills --organization '$ORG_ID' --role draft --skills outline,lesson-project,data-processing-spec,typst; rm -rf \"\$stage\""
+
+echo "[7/8] Configure Nginx and TLS"
+printf 'server { listen 80; listen [::]:80; server_name %s; client_max_body_size 2m; location / { proxy_pass http://127.0.0.1:%s; proxy_http_version 1.1; proxy_set_header Host $host; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_set_header X-Forwarded-Proto $scheme; proxy_set_header Upgrade $http_upgrade; proxy_set_header Connection $connection_upgrade; proxy_read_timeout 3600s; proxy_send_timeout 3600s; proxy_buffering off; } }\n' "$DOMAIN" "$HUB_PORT" | "${SSH[@]}" "install -o root -g root -m 0644 /dev/stdin '/etc/nginx/sites-available/$INSTANCE_ID'; ln -sfn '/etc/nginx/sites-available/$INSTANCE_ID' '/etc/nginx/sites-enabled/$INSTANCE_ID'; nginx -t; systemctl reload nginx; certbot --nginx --non-interactive --agree-tos --redirect --register-unsafely-without-email -d '$DOMAIN'"
+
+echo "[8/8] Start and verify"
+"${SSH[@]}" "systemctl enable --now '$UNIT'; systemctl is-active --quiet '$UNIT'; curl --fail --silent 'http://127.0.0.1:$HUB_PORT/api/healthz' >/dev/null; rm -f '$remote_stage/bootstrap.json'"
+curl --fail --silent --show-error "$PUBLIC_URL/api/healthz" >/dev/null
+echo "Deployed $INSTANCE_ID at $PUBLIC_URL"

hub/deploy/new_silo.sh

@@ -0,0 +1,412 @@
+#!/usr/bin/env bash
+#
+# A wizard — walks a human through a manual procedure step by step.
+# Generated by the /wizard skill.
+#
+# Everything above the "STAGES" marker is the wizard library: do not hand-edit
+# it. Author the per-step stages below the marker.
+
+set -euo pipefail
+
+# ──────────────────────────────────────────────────────────────────────────
+# Wizard library — delightful, consistent UX. Identical across every wizard.
+# ──────────────────────────────────────────────────────────────────────────
+
+if [[ -t 1 ]] && command -v tput >/dev/null 2>&1 && [[ "$(tput colors 2>/dev/null || echo 0)" -ge 8 ]]; then
+  BOLD=$(tput bold); DIM=$(tput dim); RESET=$(tput sgr0)
+  BLUE=$(tput setaf 4); GREEN=$(tput setaf 2); YELLOW=$(tput setaf 3); RED=$(tput setaf 1)
+else
+  BOLD=""; DIM=""; RESET=""; BLUE=""; GREEN=""; YELLOW=""; RED=""
+fi
+
+# Author sets these two at the top of the stages section.
+TOTAL_STAGES=0
+TOTAL_MINUTES=0
+
+_STAGE_INDEX=0
+_MINUTES_ELAPSED=0
+ENV_FILE="${ENV_FILE:-.env}"
+WRITTEN_ENV=()    # KEYs written to ENV_FILE this run
+WRITTEN_SECRET=() # secret NAMEs set this run
+SKIPPED=()        # things we couldn't do (e.g. gh missing)
+
+# _clear — wipe the terminal so only the current step is on screen. No-op when
+# output isn't a terminal, so piped logs stay readable.
+_clear() {
+  [[ -t 1 ]] || return 0
+  if command -v tput >/dev/null 2>&1; then tput clear; else printf '\033[2J\033[3J\033[H'; fi
+}
+
+# banner "Title" — opening frame: what this wizard does and how long it takes.
+banner() {
+  _clear
+  printf '\n%s%s  %s%s\n' "$BOLD" "$BLUE" "$1" "$RESET"
+  printf '%s  %s stages · about %s minutes%s\n\n' \
+    "$DIM" "$TOTAL_STAGES" "$TOTAL_MINUTES" "$RESET"
+  printf '%s  You drive the browser; this wizard tells you exactly what to do and\n' "$DIM"
+  printf '  captures the values you copy back. Stop any time with Ctrl-C and re-run\n'
+  printf '  later — it remembers values already saved.%s\n' "$RESET"
+  pause "Ready to start?"
+}
+
+# stage "Name" <minutes> — clear the screen, then announce a stage and show
+# progress + time remaining. Clearing keeps only the current step on screen.
+stage() {
+  _clear
+  _STAGE_INDEX=$((_STAGE_INDEX + 1))
+  local remaining=$((TOTAL_MINUTES - _MINUTES_ELAPSED))
+  (( remaining < 0 )) && remaining=0
+  _MINUTES_ELAPSED=$((_MINUTES_ELAPSED + ${2:-0}))
+  printf '\n%s%s▸ Stage %s/%s · %s%s  %s(~%s min left)%s\n' \
+    "$BOLD" "$BLUE" "$_STAGE_INDEX" "$TOTAL_STAGES" "$1" "$RESET" "$DIM" "$remaining" "$RESET"
+}
+
+# say "..." — a plain instruction line.
+say()  { printf '  %s\n' "$1"; }
+# step "..." — a numbered-feeling action the human takes in the browser.
+step() { printf '  %s•%s %s\n' "$BLUE" "$RESET" "$1"; }
+note() { printf '  %s%s%s\n' "$DIM" "$1" "$RESET"; }
+warn() { printf '  %s⚠ %s%s\n' "$YELLOW" "$1" "$RESET"; }
+
+# open_url URL — open in the human's browser, cross-platform incl. WSL.
+open_url() {
+  local url="$1"
+  printf '  %s↗ opening%s %s\n' "$GREEN" "$RESET" "$url"
+  { if   command -v wslview     >/dev/null 2>&1; then wslview "$url"
+    elif command -v explorer.exe >/dev/null 2>&1; then explorer.exe "$url"
+    elif command -v xdg-open    >/dev/null 2>&1; then xdg-open "$url"
+    elif command -v open        >/dev/null 2>&1; then open "$url"
+    else warn "couldn't open a browser — visit it manually: $url"; fi
+  } >/dev/null 2>&1 || warn "couldn't open a browser — visit it manually: $url"
+}
+
+# pause "msg" — wait for the human to confirm they've done the manual part.
+pause() {
+  printf '  %s%s%s ' "$DIM" "${1:-Press Enter to continue}" "$RESET"
+  read -r _ || true
+}
+
+# confirm "question" — y/N gate; returns success on yes.
+confirm() {
+  local reply=""
+  printf '  %s? %s [y/N] ' "$YELLOW" "$1"
+  read -r reply || true
+  [[ "$reply" =~ ^[Yy] ]]
+}
+
+# _existing KEY — current value of KEY in ENV_FILE, if any.
+_existing() {
+  [[ -f "$ENV_FILE" ]] || return 1
+  local line; line=$(grep -E "^${1}=" "$ENV_FILE" | tail -n1) || return 1
+  printf '%s' "${line#*=}"
+}
+
+# ask KEY "Prompt" — read a value into $KEY. Offers the existing .env value as
+# a default on re-runs (Enter keeps it). Visible input (non-secret).
+ask() {
+  local key="$1" prompt="$2" current input
+  current=$(_existing "$key" || true)
+  if [[ -n "$current" ]]; then
+    printf '  %s%s%s %s[Enter keeps current]%s ' "$BOLD" "$prompt" "$RESET" "$DIM" "$RESET"
+  else
+    printf '  %s%s%s ' "$BOLD" "$prompt" "$RESET"
+  fi
+  read -r input || true
+  [[ -z "$input" && -n "$current" ]] && input="$current"
+  printf -v "$key" '%s' "$input"
+}
+
+# ask_secret KEY "Prompt" — like ask, but input is hidden.
+ask_secret() {
+  local key="$1" prompt="$2" current input
+  current=$(_existing "$key" || true)
+  if [[ -n "$current" ]]; then
+    printf '  %s%s%s %s[Enter keeps current]%s ' "$BOLD" "$prompt" "$RESET" "$DIM" "$RESET"
+  else
+    printf '  %s%s%s ' "$BOLD" "$prompt" "$RESET"
+  fi
+  read -rs input || true
+  printf '\n'
+  [[ -z "$input" && -n "$current" ]] && input="$current"
+  printf -v "$key" '%s' "$input"
+}
+
+# write_env KEY VALUE — upsert KEY=VALUE into ENV_FILE (creates it; replaces
+# any existing line). Idempotent.
+write_env() {
+  local key="$1" value="$2" tmp
+  touch "$ENV_FILE"
+  tmp=$(mktemp)
+  grep -vE "^${key}=" "$ENV_FILE" > "$tmp" || true
+  printf '%s=%s\n' "$key" "$value" >> "$tmp"
+  mv "$tmp" "$ENV_FILE"
+  WRITTEN_ENV+=("$key")
+  printf '  %s✓ wrote%s %s → %s\n' "$GREEN" "$RESET" "$key" "$ENV_FILE"
+}
+
+# set_secret NAME VALUE — set a GitHub Actions repo secret via gh. Falls back
+# to a warning (and records it) if gh is unavailable or unauthenticated.
+set_secret() {
+  local name="$1" value="$2"
+  if command -v gh >/dev/null 2>&1 && gh auth status >/dev/null 2>&1; then
+    if printf '%s' "$value" | gh secret set "$name" >/dev/null 2>&1; then
+      WRITTEN_SECRET+=("$name")
+      printf '  %s✓ set%s GitHub secret %s\n' "$GREEN" "$RESET" "$name"
+      return
+    fi
+  fi
+  SKIPPED+=("GitHub secret $name (set it manually: gh secret set $name)")
+  warn "skipped GitHub secret $name — gh not ready; set it later"
+}
+
+# set_var NAME VALUE — set a GitHub Actions repo variable (non-secret).
+set_var() {
+  local name="$1" value="$2"
+  if command -v gh >/dev/null 2>&1 && gh auth status >/dev/null 2>&1; then
+    if gh variable set "$name" --body "$value" >/dev/null 2>&1; then
+      printf '  %s✓ set%s GitHub variable %s\n' "$GREEN" "$RESET" "$name"
+      return
+    fi
+  fi
+  SKIPPED+=("GitHub variable $name")
+  warn "skipped GitHub variable $name — gh not ready; set it later"
+}
+
+# finish — clear, then a closing summary of everything configured.
+finish() {
+  _clear
+  printf '\n%s%s  ✓ Setup complete%s\n' "$BOLD" "$GREEN" "$RESET"
+  (( ${#WRITTEN_ENV[@]} ))    && note "wrote ${#WRITTEN_ENV[@]} value(s) to $ENV_FILE: ${WRITTEN_ENV[*]}"
+  (( ${#WRITTEN_SECRET[@]} )) && note "set ${#WRITTEN_SECRET[@]} GitHub secret(s): ${WRITTEN_SECRET[*]}"
+  if (( ${#SKIPPED[@]} )); then
+    printf '\n'; warn "still to do by hand:"
+    for s in "${SKIPPED[@]}"; do note "  - $s"; done
+  fi
+  printf '\n'
+}
+
+# ──────────────────────────────────────────────────────────────────────────
+# STAGES — author this section. One stage() per step the human takes.
+# Replace the example below. Set the two totals to match the stages you write.
+# ──────────────────────────────────────────────────────────────────────────
+
+TOTAL_STAGES=7
+TOTAL_MINUTES=35
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"
+PLAN_ROOT="${CPH_SILO_PLAN_ROOT:-$HOME/.cph-silo-plans}"
+umask 077
+
+require_value() {
+  local key="$1" value="$2"
+  [[ -n "$value" ]] || { warn "$key is required"; exit 1; }
+  [[ "$value" != *$'\n'* && "$value" != *$'\r'* ]] || {
+    warn "$key must be a single line"
+    exit 1
+  }
+}
+
+seed_default() {
+  local key="$1" value="$2"
+  _existing "$key" >/dev/null 2>&1 || write_env "$key" "$value"
+}
+
+capture() {
+  local key="$1" prompt="$2"
+  ask "$key" "$prompt"
+  require_value "$key" "${!key}"
+  write_env "$key" "${!key}"
+}
+
+capture_secret() {
+  local key="$1" prompt="$2"
+  ask_secret "$key" "$prompt"
+  require_value "$key" "${!key}"
+  write_env "$key" "${!key}"
+}
+
+allocate_host_slot() {
+  local allocation local_reserved="" answers_file reserved_port
+  for answers_file in "$PLAN_ROOT"/*/answers.env; do
+    [ -f "$answers_file" ] || continue
+    reserved_port="$(sed -n 's/^HUB_PORT=//p' "$answers_file" | tail -n1)"
+    [[ "$reserved_port" =~ ^[0-9]+$ ]] || continue
+    local_reserved="${local_reserved:+$local_reserved,}$reserved_port"
+  done
+  allocation="$(ssh \
+    -i "$DEPLOY_SSH_KEY" \
+    -p "$DEPLOY_SSH_PORT" \
+    -o BatchMode=yes \
+    -o StrictHostKeyChecking=accept-new \
+    "$DEPLOY_USER@$DEPLOY_HOST" \
+    bash -s -- "$local_reserved" <<'REMOTE'
+      set -euo pipefail
+      local_reserved=",${1:-},"
+      for candidate in $(seq 8788 8999); do
+        reserved=false
+        if [[ "$local_reserved" == *",$candidate,"* ]]; then
+          continue
+        fi
+        for env_file in /srv/curriculum-project-hub/.secrets/*/platform.env; do
+          [ -f "$env_file" ] || continue
+          if [ "$(sed -n "s/^PORT=//p" "$env_file")" = "$candidate" ]; then
+            reserved=true
+            break
+          fi
+        done
+        workspace="/w/$candidate"
+        if [ "$reserved" = false ] && ! ss -H -ltn "sport = :$candidate" | grep -q . && [ ! -e "$workspace" ]; then
+          printf "%s %s\n" "$candidate" "$workspace"
+          exit 0
+        fi
+      done
+      echo "no free Alpha Silo slot in 8788..8999" >&2
+      exit 1
+REMOTE
+  )"
+  read -r HUB_PORT WORKSPACE_ROOT <<<"$allocation"
+  require_value HUB_PORT "$HUB_PORT"
+  require_value WORKSPACE_ROOT "$WORKSPACE_ROOT"
+  write_env HUB_PORT "$HUB_PORT"
+  write_env WORKSPACE_ROOT "$WORKSPACE_ROOT"
+}
+
+banner "New Alpha Silo"
+
+stage "Organization identity and private plan" 3
+say "One run creates one Organization's private deployment bundle."
+ask ORGANIZATION_SLUG "Organization slug (lowercase, max 24 chars; e.g. school-a):"
+require_value ORGANIZATION_SLUG "$ORGANIZATION_SLUG"
+[[ "$ORGANIZATION_SLUG" =~ ^[a-z0-9]([a-z0-9-]{0,22}[a-z0-9])?$ ]] || {
+  warn "invalid Organization slug"
+  exit 1
+}
+INSTANCE_ID="$ORGANIZATION_SLUG"
+ORGANIZATION_ID="$ORGANIZATION_SLUG"
+OUTPUT_DIR="$PLAN_ROOT/$INSTANCE_ID"
+mkdir -p "$OUTPUT_DIR"
+chmod 0700 "$OUTPUT_DIR"
+ENV_FILE="$OUTPUT_DIR/answers.env"
+touch "$ENV_FILE"
+chmod 0600 "$ENV_FILE"
+write_env INSTANCE_ID "$INSTANCE_ID"
+write_env ORGANIZATION_ID "$ORGANIZATION_ID"
+write_env ORGANIZATION_SLUG "$ORGANIZATION_SLUG"
+capture ORGANIZATION_NAME "Organization display name:"
+
+stage "Host, release, and isolation" 5
+say "The Alpha host is platform-managed. Port and short workspace path are allocated from live host state."
+DEPLOY_HOST="${CPH_ALPHA_HOST:-39.107.254.4}"
+DEPLOY_USER="${CPH_ALPHA_DEPLOY_USER:-root}"
+DEPLOY_SSH_PORT="${CPH_ALPHA_SSH_PORT:-22}"
+write_env DEPLOY_HOST "$DEPLOY_HOST"
+write_env DEPLOY_USER "$DEPLOY_USER"
+write_env DEPLOY_SSH_PORT "$DEPLOY_SSH_PORT"
+seed_default DEPLOY_BASE "/srv/curriculum-project-hub"
+seed_default DEPLOY_SSH_KEY "$HOME/.ssh/id_ed25519"
+write_env RELEASE_ID "v$(node -p 'require(process.argv[1]).version' "$REPO_ROOT/hub/package.json")"
+seed_default MEMORY_MAX "16G"
+seed_default CPU_QUOTA "400%"
+seed_default TASKS_MAX "512"
+seed_default CPH_BIN "/usr/local/bin/cph"
+note "Managed Alpha host: $DEPLOY_USER@$DEPLOY_HOST:$DEPLOY_SSH_PORT"
+DEPLOY_SSH_KEY="$(_existing DEPLOY_SSH_KEY)"
+MEMORY_MAX="$(_existing MEMORY_MAX)"
+CPU_QUOTA="$(_existing CPU_QUOTA)"
+TASKS_MAX="$(_existing TASKS_MAX)"
+[ -f "$DEPLOY_SSH_KEY" ] || { warn "managed SSH key is missing: $DEPLOY_SSH_KEY"; exit 1; }
+if [[ "$(_existing HUB_PORT || true)" =~ ^(878[8-9]|87[9][0-9]|8[89][0-9]{2})$ ]] && \
+   [ "$(_existing WORKSPACE_ROOT || true)" = "/w/$(_existing HUB_PORT)" ]; then
+  HUB_PORT="$(_existing HUB_PORT)"
+  WORKSPACE_ROOT="$(_existing WORKSPACE_ROOT)"
+  note "Keeping allocated host slot: port $HUB_PORT, workspace $WORKSPACE_ROOT"
+else
+  say "Checking existing Silo environments, listening sockets, and workspace paths..."
+  allocate_host_slot
+  note "Allocated host slot: port $HUB_PORT, workspace $WORKSPACE_ROOT"
+fi
+note "Platform ceilings: MemoryMax=$MEMORY_MAX, CPUQuota=$CPU_QUOTA, TasksMax=$TASKS_MAX"
+
+stage "Dedicated PostgreSQL database" 4
+say "A PostgreSQL server may be shared, but this Silo gets a distinct login role and database."
+seed_default DATABASE_HOST "127.0.0.1"
+seed_default DATABASE_PORT "5432"
+seed_default DATABASE_NAME "cph_${INSTANCE_ID//-/_}"
+seed_default DATABASE_USER "cph_${INSTANCE_ID//-/_}"
+DATABASE_NAME="$(_existing DATABASE_NAME)"
+if ! _existing DATABASE_PASSWORD >/dev/null 2>&1; then
+  DATABASE_PASSWORD="$(openssl rand -base64 36 | tr -d '\n')"
+  write_env DATABASE_PASSWORD "$DATABASE_PASSWORD"
+fi
+note "The generated OPERATE.md uses an interactive/protected SQL path; the password is never put in a command argument."
+
+stage "Public URL and Feishu app" 9
+open_url "https://open.feishu.cn/app"
+say "Create or open the Organization's own app. Copy credentials from Credentials & Basic Info."
+PUBLIC_BASE_URL="https://${ORGANIZATION_SLUG}.${CPH_ALPHA_BASE_DOMAIN:-educraft.paradigm-edu.net}"
+write_env PUBLIC_BASE_URL "$PUBLIC_BASE_URL"
+note "Platform-assigned public URL: $PUBLIC_BASE_URL"
+capture FEISHU_APP_ID "Feishu App ID:"
+capture_secret FEISHU_APP_SECRET "Feishu App Secret:"
+say "Resolving the bot Open ID from Feishu..."
+FEISHU_BOT_OPEN_ID="$(printf '%s\0%s\0' "$FEISHU_APP_ID" "$FEISHU_APP_SECRET" | node "$SCRIPT_DIR/resolve_feishu_bot.mjs")"
+write_env FEISHU_BOT_OPEN_ID "$FEISHU_BOT_OPEN_ID"
+note "Resolved bot identity: $FEISHU_BOT_OPEN_ID"
+open_url "https://open.feishu.cn/document/server-docs/contact-v3/user/get"
+step "In the user/get page, click the user_id value picker, select the first OWNER, and copy the returned open_id."
+capture OWNER_OPEN_ID "OWNER Open ID (ou_...):"
+capture OWNER_DISPLAY_NAME "OWNER display name:"
+write_env OWNER_UNION_ID ""
+say "The exact redirect URL and acceptance steps will be written to OPERATE.md."
+
+stage "Provider and Alpha limits" 5
+say "Use a provider credential exclusive to this Organization. Host proxy setup is a separate prerequisite."
+seed_default PROVIDER_ID "openrouter"
+seed_default PROVIDER_BASE_URL "https://openrouter.ai/api"
+seed_default DEFAULT_MODEL "anthropic/claude-sonnet-5"
+seed_default DEFAULT_ROLE_ID "draft"
+seed_default DEFAULT_ROLE_LABEL "智能助手"
+seed_default MAX_TURNS "25"
+seed_default MAX_CONCURRENT_RUNS "4"
+seed_default MAX_RUN_SECONDS "900"
+seed_default HTTP_BODY_LIMIT_BYTES "1048576"
+seed_default MAX_FILES_PER_MESSAGE "8"
+seed_default MAX_FILE_BYTES "26214400"
+seed_default HTTP_REQUESTS_PER_MINUTE "120"
+seed_default FEISHU_EVENTS_PER_MINUTE "120"
+capture_secret PROVIDER_AUTH_TOKEN "Provider auth token:"
+for key in PROVIDER_ID PROVIDER_BASE_URL DEFAULT_MODEL DEFAULT_ROLE_ID DEFAULT_ROLE_LABEL \
+  MAX_TURNS MAX_CONCURRENT_RUNS MAX_RUN_SECONDS HTTP_BODY_LIMIT_BYTES \
+  MAX_FILES_PER_MESSAGE MAX_FILE_BYTES HTTP_REQUESTS_PER_MINUTE FEISHU_EVENTS_PER_MINUTE; do
+  printf -v "$key" '%s' "$(_existing "$key")"
+done
+write_env APPROVED_SKILLS "outline,lesson-project,data-processing-spec,typst"
+note "Platform runtime: OpenRouter, concurrency 4, default education role, curated skills."
+if ! _existing HUB_SESSION_SECRET >/dev/null 2>&1; then
+  command -v openssl >/dev/null 2>&1 || { warn "openssl is required"; exit 1; }
+  HUB_SESSION_SECRET="$(openssl rand -hex 32)"
+  write_env HUB_SESSION_SECRET "$HUB_SESSION_SECRET"
+fi
+
+stage "Render the private deployment bundle" 2
+say "This renders platform.env, bootstrap.json, deploy.env, nginx.conf and OPERATE.md."
+command -v node >/dev/null 2>&1 || { warn "Node.js is required to render safely"; exit 1; }
+node "$SCRIPT_DIR/render_new_silo_bundle.mjs" "$ENV_FILE" "$OUTPUT_DIR"
+chmod 0700 "$OUTPUT_DIR"
+chmod 0600 "$OUTPUT_DIR"/*
+say "Bundle: $OUTPUT_DIR"
+warn "It contains database, Feishu, provider and session secrets. Never commit or paste it."
+
+stage "Operator handoff and gates" 7
+say "The deployment package is an internal retry/audit checkpoint; you do not operate it manually."
+step "Target: $PUBLIC_BASE_URL → $DEPLOY_HOST:$HUB_PORT"
+step "Resources: database $DATABASE_NAME, workspace $WORKSPACE_ROOT, service cph-hub-$INSTANCE_ID"
+warn "Confirmation will create server, database, TLS, and runtime state."
+if confirm "Deploy this Organization now?"; then
+  bash "$SCRIPT_DIR/apply_new_silo.sh" "$OUTPUT_DIR"
+else
+  warn "deployment skipped; rerun the wizard later and keep existing answers"
+fi
+
+finish

hub/deploy/render_new_silo_bundle.mjs

@@ -0,0 +1,215 @@
+#!/usr/bin/env node
+
+import { chmod, mkdir, readFile, writeFile } from "node:fs/promises";
+import { resolve } from "node:path";
+
+const [answersPath, outputPath] = process.argv.slice(2);
+if (!answersPath || !outputPath) {
+  throw new Error("usage: render_new_silo_bundle.mjs ANSWERS_ENV OUTPUT_DIR");
+}
+
+function parseAnswers(source) {
+  const values = new Map();
+  for (const [index, line] of source.split("\n").entries()) {
+    if (!line || line.startsWith("#")) continue;
+    const separator = line.indexOf("=");
+    if (separator < 1) throw new Error(`invalid answers line ${index + 1}`);
+    const key = line.slice(0, separator);
+    const value = line.slice(separator + 1);
+    if (!/^[A-Z][A-Z0-9_]*$/.test(key)) {
+      throw new Error(`invalid answers key on line ${index + 1}: ${key}`);
+    }
+    if (value.includes("\r") || value.includes("\n")) {
+      throw new Error(`multiline value is not supported: ${key}`);
+    }
+    values.set(key, value);
+  }
+  return values;
+}
+
+function required(values, key) {
+  const value = values.get(key);
+  if (!value) throw new Error(`missing required answer: ${key}`);
+  return value;
+}
+
+function optional(values, key, fallback = "") {
+  return values.get(key) || fallback;
+}
+
+function assertMatch(label, value, pattern) {
+  if (!pattern.test(value)) throw new Error(`invalid ${label}: ${value}`);
+}
+
+function envLine(key, value) {
+  if (/[\r\n]/.test(value)) throw new Error(`unsafe newline in ${key}`);
+  return `${key}=${value}`;
+}
+
+const answers = parseAnswers(await readFile(resolve(answersPath), "utf8"));
+const instanceId = required(answers, "INSTANCE_ID");
+const orgId = required(answers, "ORGANIZATION_ID");
+const orgSlug = required(answers, "ORGANIZATION_SLUG");
+const port = required(answers, "HUB_PORT");
+const workspaceRoot = required(answers, "WORKSPACE_ROOT");
+const publicBaseUrl = required(answers, "PUBLIC_BASE_URL").replace(/\/$/, "");
+const domain = new URL(publicBaseUrl).hostname;
+const databasePassword = required(answers, "DATABASE_PASSWORD");
+const databaseUrl = `postgresql://${encodeURIComponent(required(answers, "DATABASE_USER"))}:${encodeURIComponent(databasePassword)}@${required(answers, "DATABASE_HOST")}:${required(answers, "DATABASE_PORT")}/${encodeURIComponent(required(answers, "DATABASE_NAME"))}`;
+
+assertMatch("INSTANCE_ID", instanceId, /^[a-z0-9](?:[a-z0-9-]{0,22}[a-z0-9])?$/);
+assertMatch("organization slug", orgSlug, /^[a-z0-9](?:[a-z0-9-]*[a-z0-9])?$/);
+assertMatch("port", port, /^[1-9][0-9]{1,4}$/);
+assertMatch("database name", required(answers, "DATABASE_NAME"), /^[a-z_][a-z0-9_]*$/);
+assertMatch("database user", required(answers, "DATABASE_USER"), /^[a-z_][a-z0-9_]*$/);
+assertMatch("default role id", required(answers, "DEFAULT_ROLE_ID"), /^[a-z][a-z0-9-]*$/);
+assertMatch("release id", required(answers, "RELEASE_ID"), /^[A-Za-z0-9._-]+$/);
+if (!publicBaseUrl.startsWith("https://")) throw new Error("PUBLIC_BASE_URL must use https");
+for (const [label, value] of [
+  ["WORKSPACE_ROOT", workspaceRoot],
+  ["DEPLOY_BASE", required(answers, "DEPLOY_BASE")],
+  ["DEPLOY_SSH_KEY", required(answers, "DEPLOY_SSH_KEY")],
+  ["CPH_BIN", required(answers, "CPH_BIN")],
+]) {
+  if (!value.startsWith("/") || /\s/.test(value)) {
+    throw new Error(`${label} must be an absolute path without whitespace: ${value}`);
+  }
+}
+if (Number(port) > 65535) throw new Error(`invalid port: ${port}`);
+if (Buffer.byteLength(workspaceRoot) > 16) {
+  throw new Error(`WORKSPACE_ROOT exceeds the 16-byte sandbox socket limit: ${workspaceRoot}`);
+}
+
+const outputDir = resolve(outputPath);
+await mkdir(outputDir, { recursive: true, mode: 0o700 });
+await chmod(outputDir, 0o700);
+
+const base = required(answers, "DEPLOY_BASE");
+const release = required(answers, "RELEASE_ID");
+const releaseHub = `${base}/releases/${release}/hub`;
+const secretDir = `${base}/.secrets/${instanceId}`;
+const envPath = `${secretDir}/platform.env`;
+const keyringPath = `${secretDir}/secret-keyring.json`;
+const unit = `cph-hub-${instanceId}.service`;
+
+const platformEnv = [
+  "# Generated by hub/deploy/new_silo.sh. Install root:root mode 0600.",
+  envLine("NODE_ENV", "production"),
+  envLine("DATABASE_URL", databaseUrl),
+  envLine("HUB_SILO_ORGANIZATION_ID", orgId),
+  envLine("HUB_SYSTEMD_UNIT", unit),
+  envLine("CPH_BIN", required(answers, "CPH_BIN")),
+  envLine("HOST", "127.0.0.1"),
+  envLine("PORT", port),
+  envLine("HUB_PROJECT_WORKSPACE_ROOT", workspaceRoot),
+  envLine("HUB_PUBLIC_BASE_URL", publicBaseUrl),
+  envLine("HUB_SESSION_SECRET", required(answers, "HUB_SESSION_SECRET")),
+  envLine("HUB_AGENT_MAX_TURNS", required(answers, "MAX_TURNS")),
+  envLine("HUB_AGENT_MAX_CONCURRENT_RUNS", required(answers, "MAX_CONCURRENT_RUNS")),
+  envLine("HUB_AGENT_MAX_RUN_SECONDS", required(answers, "MAX_RUN_SECONDS")),
+  envLine("HUB_HTTP_BODY_LIMIT_BYTES", required(answers, "HTTP_BODY_LIMIT_BYTES")),
+  envLine("HUB_MAX_FILES_PER_MESSAGE", required(answers, "MAX_FILES_PER_MESSAGE")),
+  envLine("HUB_MAX_FILE_BYTES", required(answers, "MAX_FILE_BYTES")),
+  envLine("HUB_HTTP_REQUESTS_PER_MINUTE", required(answers, "HTTP_REQUESTS_PER_MINUTE")),
+  envLine("HUB_FEISHU_EVENTS_PER_MINUTE", required(answers, "FEISHU_EVENTS_PER_MINUTE")),
+  envLine("HUB_FEISHU_LISTENER_ENABLED", "true"),
+  envLine("HTTP_PROXY", "http://127.0.0.1:7890"),
+  envLine("HTTPS_PROXY", "http://127.0.0.1:7890"),
+  envLine("ALL_PROXY", "socks5h://127.0.0.1:7890"),
+  envLine("NO_PROXY", "127.0.0.1,localhost,::1"),
+  envLine("CPH_SANDBOX_EXTRA_DENY_READ", `${envPath}:${keyringPath}`),
+  "",
+].join("\n");
+
+const bootstrap = {
+  organization: {
+    id: orgId,
+    slug: orgSlug,
+    name: required(answers, "ORGANIZATION_NAME"),
+  },
+  owner: {
+    openId: required(answers, "OWNER_OPEN_ID"),
+    displayName: required(answers, "OWNER_DISPLAY_NAME"),
+  },
+  feishu: {
+    appId: required(answers, "FEISHU_APP_ID"),
+    appSecret: required(answers, "FEISHU_APP_SECRET"),
+    botOpenId: required(answers, "FEISHU_BOT_OPEN_ID"),
+  },
+  provider: {
+    providerId: required(answers, "PROVIDER_ID"),
+    baseUrl: required(answers, "PROVIDER_BASE_URL"),
+    authToken: required(answers, "PROVIDER_AUTH_TOKEN"),
+  },
+  teams: [],
+};
+const ownerUnionId = optional(answers, "OWNER_UNION_ID");
+if (ownerUnionId) bootstrap.owner.unionId = ownerUnionId;
+
+const deployEnv = [
+  "# Source this file locally before deploy_platform.sh (contains no app/provider secrets).",
+  envLine("PLATFORM_DEPLOY_HOST", required(answers, "DEPLOY_HOST")),
+  envLine("PLATFORM_DEPLOY_SSH_KEY", required(answers, "DEPLOY_SSH_KEY")),
+  envLine("PLATFORM_DEPLOY_USER", required(answers, "DEPLOY_USER")),
+  envLine("PLATFORM_DEPLOY_PORT", required(answers, "DEPLOY_SSH_PORT")),
+  envLine("PLATFORM_DEPLOY_HUB_PORT", port),
+  envLine("PLATFORM_DEPLOY_BASE", base),
+  envLine("PLATFORM_DEPLOY_RELEASE", release),
+  envLine("PLATFORM_DEPLOY_INSTANCE", instanceId),
+  envLine("PLATFORM_DEPLOY_WORKSPACE_ROOT", workspaceRoot),
+  envLine("PLATFORM_DEPLOY_MEMORY_MAX", required(answers, "MEMORY_MAX")),
+  envLine("PLATFORM_DEPLOY_CPU_QUOTA", required(answers, "CPU_QUOTA")),
+  envLine("PLATFORM_DEPLOY_TASKS_MAX", required(answers, "TASKS_MAX")),
+  envLine("PLATFORM_DEPLOY_HEALTH_URL", `http://127.0.0.1:${port}/api/healthz`),
+  "",
+].join("\n");
+
+const nginx = `# Install as /etc/nginx/conf.d/${instanceId}.conf after obtaining TLS certificates.\nserver {\n  listen 80;\n  server_name ${domain};\n  return 301 https://$host$request_uri;\n}\n\nserver {\n  listen 443 ssl http2;\n  server_name ${domain};\n\n  ssl_certificate /etc/letsencrypt/live/${domain}/fullchain.pem;\n  ssl_certificate_key /etc/letsencrypt/live/${domain}/privkey.pem;\n\n  location / {\n    proxy_pass http://127.0.0.1:${port};\n    proxy_http_version 1.1;\n    proxy_set_header Host $host;\n    proxy_set_header X-Real-IP $remote_addr;\n    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;\n    proxy_set_header X-Forwarded-Proto https;\n  }\n}\n`;
+
+const defaultRolePrompt = `你是一位教育机构智能助手，服务于学校和教培机构的日常管理与协作。
+## 核心能力
+**教务管理**：熟悉排课调课、班级管理、课时统计、师资调度等常见管理流程。
+**教研支持**：理解课程设计、教案编写、例题/变式题、定理证明、随堂练习、阶段性测验、知识点拆解、大纲对标等教学业务概念。
+**数字化操作**：熟练使用shell命令行进行文件管理、批量处理、脚本编写、文本处理（grep/sed/awk等）、自动化任务；能协助处理Markdown、Typst等文档、表格数据、格式转换等。
+## 行为准则
+- **实事求是**：不确定时明确说明，能力边界外的任务如实告知，不为完成目标而编造信息或勉强输出
+- **简洁优先**：直接给出答案或方案，省略铺垫和客套
+- **按需展开**：仅在问题复杂或用户明确要求时提供详细说明
+- **结构清晰**：多步骤内容使用编号，便于飞书阅读
+- **可操作性**：涉及操作时给出具体命令或步骤
+保持专业、高效，像一位熟悉业务的同事。
+`;
+
+const operate = `# ${instanceId} deployment commands\n\nThis file contains no application/provider secrets. Run commands deliberately; do not source answers.env.\n\n## 1. DNS and Feishu\n\n- Point \`${domain}\` to \`${required(answers, "DEPLOY_HOST")}\`.\n- In the Feishu app, configure redirect URL: \`${publicBaseUrl}/auth/feishu/callback\`.\n- Enable the required bot/message/contact permissions and event subscription described in the customer setup document.\n\n## 2. Database (run as the PostgreSQL administrator)\n\nCreate one role and one database for this Silo. The password is in \`answers.env\`; use an interactive client or a protected SQL file, never a command-line argument.\n\n\`\`\`sql\nCREATE ROLE ${required(answers, "DATABASE_USER")} LOGIN PASSWORD '<copy DATABASE_PASSWORD from answers.env>';\nCREATE DATABASE ${required(answers, "DATABASE_NAME")} OWNER ${required(answers, "DATABASE_USER")};\n\`\`\`\n\n## 3. Publish the immutable release\n\nThe normal deploy script seeds the first-instance secrets and exits 78. That exit is expected only on this first pass.\n\n\`\`\`sh\nset -a; . ./deploy.env; set +a\nbash hub/deploy/deploy_platform.sh\n\`\`\`\n\n## 4. Install the generated secrets\n\nCopy \`platform.env\` and \`bootstrap.json\` to the server through a protected channel. On the server:\n\n\`\`\`sh\ninstall -d -o root -g root -m 0700 '${secretDir}'\ninstall -o root -g root -m 0600 platform.env '${envPath}'\ninstall -o root -g root -m 0600 bootstrap.json '/root/${instanceId}-bootstrap.json'\n# Copy ${keyringPath} to separate off-host recovery storage before continuing.\n\`\`\`\n\n## 5. Migrate, install the stopped service, and bootstrap\n\n\`\`\`sh\nset -a; . '${envPath}'; set +a\nnode '${releaseHub}/node_modules/prisma/build/index.js' migrate deploy --schema '${releaseHub}/prisma/schema.prisma'\nBASE='${base}' HUB_DIR='${releaseHub}' INSTANCE_ID='${instanceId}' WORKSPACE_ROOT='${workspaceRoot}' PORT='${port}' MEMORY_MAX='${required(answers, "MEMORY_MAX")}' CPU_QUOTA='${required(answers, "CPU_QUOTA")}' TASKS_MAX='${required(answers, "TASKS_MAX")}' bash '${releaseHub}/deploy/install_service.sh'\nnode '${releaseHub}/dist/deployment/bootstrap-silo-cli.js' --config-file '/root/${instanceId}-bootstrap.json' --keyring-file '${keyringPath}'\nrm -f '/root/${instanceId}-bootstrap.json'\n\`\`\`\n\n## 6. Runtime role and skills\n\nRuntime role/skill configuration is intentionally separate from the release. Follow NEW_SILO_RUNBOOK.md, staging each approved skill directory outside the immutable release, then use \`agent_config.sh\`. Do not silently copy a local personal skill collection.\n\n## 7. Nginx, start, and acceptance\n\nInstall \`nginx.conf\`, run \`nginx -t\`, reload Nginx, then:\n\n\`\`\`sh\nsystemctl start '${unit}'\nsystemctl is-active '${unit}'\ncurl --fail --silent --show-error 'http://127.0.0.1:${port}/api/healthz'\ncurl --fail --silent --show-error '${publicBaseUrl}/api/healthz'\njournalctl -u '${unit}' --since '-10 min' --no-pager\n\`\`\`\n\nAcceptance requires: Feishu OAuth completes, OWNER and a non-OWNER can @bot, a second group creates a separate project/session, Typst works when that skill is enabled, and restart preserves session cursor. Then take the first off-host backup.\n\n## 8. Later releases\n\nAfter the instance exists, source \`deploy.env\` with the new release id and run \`deploy_platform.sh\`. Never reuse another org's database, secret directory, workspace root, port, service identity, domain, Feishu app, or provider credential.\n`;
+
+const selectedSkills = optional(answers, "APPROVED_SKILLS");
+const roleInstructions = `
+## Appendix: default runtime role
+
+Copy \`default-role-prompt.md\` to \`/root/${instanceId}-default-role-prompt.md\`.
+After installing each reviewed skill with \`agent_config.sh install-skill\`, run:
+
+\`\`\`sh
+INSTANCE_ID=${JSON.stringify(instanceId)} ENV_FILE=${JSON.stringify(envPath)} HUB_DIR=${JSON.stringify(releaseHub)} bash ${JSON.stringify(`${releaseHub}/deploy/agent_config.sh`)} upsert-role --organization ${JSON.stringify(orgId)} --role ${JSON.stringify(required(answers, "DEFAULT_ROLE_ID"))} --label ${JSON.stringify(required(answers, "DEFAULT_ROLE_LABEL"))} --model ${JSON.stringify(required(answers, "DEFAULT_MODEL"))} --system-prompt-file ${JSON.stringify(`/root/${instanceId}-default-role-prompt.md`)} --tools-json null
+${selectedSkills ? `INSTANCE_ID=${JSON.stringify(instanceId)} ENV_FILE=${JSON.stringify(envPath)} HUB_DIR=${JSON.stringify(releaseHub)} bash ${JSON.stringify(`${releaseHub}/deploy/agent_config.sh`)} set-role-skills --organization ${JSON.stringify(orgId)} --role ${JSON.stringify(required(answers, "DEFAULT_ROLE_ID"))} --skills ${JSON.stringify(selectedSkills)}
+` : "# No skills selected in the wizard; set-role-skills remains an explicit operator step.\n"}INSTANCE_ID=${JSON.stringify(instanceId)} ENV_FILE=${JSON.stringify(envPath)} HUB_DIR=${JSON.stringify(releaseHub)} bash ${JSON.stringify(`${releaseHub}/deploy/agent_config.sh`)} list --organization ${JSON.stringify(orgId)}
+\`\`\`
+
+Changing a role's model, system prompt, tools or skill selection archives that
+role's existing sessions by design. Finish this setup before inviting Alpha users.
+`;
+
+async function privateWrite(name, contents) {
+  const path = resolve(outputDir, name);
+  await writeFile(path, contents, { encoding: "utf8", mode: 0o600 });
+  await chmod(path, 0o600);
+}
+
+await privateWrite("platform.env", platformEnv);
+await privateWrite("bootstrap.json", `${JSON.stringify(bootstrap, null, 2)}\n`);
+await privateWrite("deploy.env", deployEnv);
+await privateWrite("nginx.conf", nginx);
+await privateWrite("default-role-prompt.md", defaultRolePrompt);
+await privateWrite("OPERATE.md", `${operate}${roleInstructions}`);
+
+console.log(`Rendered private Silo bundle: ${outputDir}`);

hub/deploy/resolve_feishu_bot.mjs

@@ -0,0 +1,27 @@
+#!/usr/bin/env node
+
+const chunks = [];
+for await (const chunk of process.stdin) chunks.push(chunk);
+const [appId, appSecret] = Buffer.concat(chunks).toString("utf8").split("\0");
+if (!appId || !appSecret) throw new Error("Feishu App ID and App Secret are required on stdin");
+
+const tokenResponse = await fetch("https://open.feishu.cn/open-apis/auth/v3/tenant_access_token/internal", {
+  method: "POST",
+  headers: { "content-type": "application/json; charset=utf-8" },
+  body: JSON.stringify({ app_id: appId, app_secret: appSecret }),
+});
+if (!tokenResponse.ok) throw new Error(`Feishu token request failed: HTTP ${tokenResponse.status}`);
+const tokenPayload = await tokenResponse.json();
+if (tokenPayload.code !== 0 || typeof tokenPayload.tenant_access_token !== "string") {
+  throw new Error(`Feishu token request failed: ${JSON.stringify(tokenPayload)}`);
+}
+
+const botResponse = await fetch("https://open.feishu.cn/open-apis/bot/v3/info", {
+  headers: { authorization: `Bearer ${tokenPayload.tenant_access_token}` },
+});
+if (!botResponse.ok) throw new Error(`Feishu bot info request failed: HTTP ${botResponse.status}`);
+const botPayload = await botResponse.json();
+if (botPayload.code !== 0 || typeof botPayload.bot?.open_id !== "string" || !botPayload.bot.open_id) {
+  throw new Error(`Feishu bot info request failed: ${JSON.stringify(botPayload)}`);
+}
+process.stdout.write(botPayload.bot.open_id);

hub/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@paradigm/hub",
-  "version": "0.0.13",
+  "version": "0.0.18",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@paradigm/hub",
-      "version": "0.0.13",
+      "version": "0.0.18",
       "dependencies": {
         "@anthropic-ai/claude-agent-sdk": "^0.3.202",
         "@fastify/cookie": "^11.0.2",

hub/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@paradigm/hub",
-  "version": "0.0.13",
+  "version": "0.0.18",
   "private": true,
   "type": "module",
   "engines": {

hub/src/agent/security.ts

@@ -23,11 +23,6 @@ const SAFE_HOST_ENV_KEYS = [
   "CPH_BIN",
 ] as const;
 
-const SANDBOX_HIDDEN_ENV_KEYS = [
-  "ANTHROPIC_AUTH_TOKEN",
-  "ANTHROPIC_API_KEY",
-] as const;
-
 // Linux sockaddr_un.sun_path is 108 bytes including the terminator. Claude's
 // sandbox appends its own user directory and randomized bridge socket names,
 // so keep our prefix well below that hard limit.
@@ -156,7 +151,9 @@ export async function createAgentSecurityPolicy(input: AgentSecurityInput): Prom
       },
       credentials: {
         files: sensitiveReadPaths.map((path) => ({ path, mode: "deny" as const })),
-        envVars: SANDBOX_HIDDEN_ENV_KEYS.map((name) => ({ name, mode: "deny" as const })),
+        // These values are short-lived loopback capabilities, not Organization
+        // provider secrets. Denying them also strips Claude's own request auth.
+        envVars: [],
       },
     },
   };

hub/src/connections/providerProxy.ts

@@ -41,6 +41,11 @@ export interface ProviderProxyDiagnostic {
     | "provider_proxy_redirect_refused"
     | "provider_proxy_upstream_failed";
   readonly category: NetworkFailureCategory | "authorization" | "request_limit" | "redirect";
+  readonly authShape?: {
+    readonly bearerLength: number;
+    readonly apiKeyLength: number;
+    readonly expectedLength: number;
+  };
 }
 
 export interface ProviderProxyOptions {
@@ -81,7 +86,7 @@ export async function openProviderProxyLease(
     sdkEnv: {
       ANTHROPIC_BASE_URL: `http://127.0.0.1:${address.port}`,
       ANTHROPIC_AUTH_TOKEN: capability,
-      ANTHROPIC_API_KEY: "",
+      ANTHROPIC_API_KEY: capability,
     },
     sensitiveValues: [capability],
     async close(): Promise<void> {
@@ -102,8 +107,18 @@ async function forwardProviderRequest(
   upstream: ProviderUpstreamCredential,
   options: ProviderProxyOptions,
 ): Promise<void> {
-  if (!authorized(request.headers.authorization, capability)) {
-    options.onDiagnostic?.({ code: "provider_proxy_unauthorized", category: "authorization" });
+  if (!authorized(request.headers.authorization, header(request.headers["x-api-key"]), capability)) {
+    options.onDiagnostic?.({
+      code: "provider_proxy_unauthorized",
+      category: "authorization",
+      authShape: {
+        bearerLength: request.headers.authorization?.startsWith("Bearer ")
+          ? request.headers.authorization.length - "Bearer ".length
+          : 0,
+        apiKeyLength: header(request.headers["x-api-key"])?.length ?? 0,
+        expectedLength: capability.length,
+      },
+    });
     response.writeHead(401, { "content-type": "text/plain; charset=utf-8" });
     response.end("unauthorized");
     return;
@@ -165,13 +180,24 @@ async function forwardProviderRequest(
   }
 }
 
-function authorized(value: string | undefined, capability: string): boolean {
-  if (value === undefined || !value.startsWith("Bearer ")) return false;
-  const supplied = Buffer.from(value.slice("Bearer ".length));
+function authorized(
+  authorization: string | undefined,
+  apiKey: string | undefined,
+  capability: string,
+): boolean {
+  const value = authorization?.startsWith("Bearer ")
+    ? authorization.slice("Bearer ".length)
+    : apiKey;
+  if (value === undefined) return false;
+  const supplied = Buffer.from(value);
   const expected = Buffer.from(capability);
   return supplied.byteLength === expected.byteLength && timingSafeEqual(supplied, expected);
 }
 
+function header(value: string | string[] | undefined): string | undefined {
+  return Array.isArray(value) ? value[0] : value;
+}
+
 function forwardedRequestHeaders(source: IncomingHttpHeaders): Headers {
   const result = new Headers();
   for (const [name, value] of Object.entries(source)) {

hub/src/feishu/trigger.ts

@@ -35,7 +35,6 @@ import type { RuntimeSettings } from "../settings/runtime.js";
 import { currentLockRunId, releaseLock } from "../lock.js";
 import { createPermissionAuthorizer, type AuthorizationDecision, type PermissionAuthorizer } from "../permission.js";
 import { writeAudit } from "../audit.js";
-import { formatRunCostLine } from "../agent/cost.js";
 import { createAgentSdkStderrSink } from "../agent/diagnostics.js";
 import { InactiveOrganizationError, lockActiveOrganization } from "../org/status.js";
 import { StreamingAgentCard } from "./card/streaming-card.js";
@@ -550,7 +549,7 @@ export function makeTriggerHandler(deps: TriggerDeps): TriggerHandler {
             : result.status === "failed" && result.error !== undefined
               ? `\u5904\u7406\u5931\u8D25: ${result.error}`
               : result.text;
-        await card.finish(finalText, { interrupted, footerText: formatRunCostLine(result.costUsd) });
+        await card.finish(finalText, { interrupted });
         const metadataPatch = sessionMetadataPatch(result.sdkSessionId);
         if (metadataPatch !== null) {
           await deps.prisma.agentSession.update({

hub/src/hub.ts

@@ -73,6 +73,7 @@ export async function startHub(): Promise<void> {
           providerId: context.providerId,
           errorCode: diagnostic.code,
           failureCategory: diagnostic.category,
+          authShape: diagnostic.authShape,
         }, "provider proxy diagnostic");
       },
     }),

hub/test/integration/trigger.test.ts

@@ -168,7 +168,8 @@ describe("trigger full lifecycle (integration)", () => {
       msgType: "interactive",
       replyInThread: undefined,
     });
-    expect(rt.sentTexts.some((text) => text.includes("mock response") && text.includes("本次成本: $0.0023"))).toBe(true);
+    expect(rt.sentTexts.some((text) => text.includes("mock response"))).toBe(true);
+    expect(rt.sentTexts.some((text) => text.includes("本次成本"))).toBe(false);
     expect(runAgentCalls).toHaveLength(1);
     expect(runAgentCalls[0]?.providerProxyEnv).toMatchObject({
       ANTHROPIC_BASE_URL: "http://127.0.0.1:12345",

hub/test/unit/agent-security.test.ts

@@ -66,10 +66,7 @@ describe("agent subprocess security policy", () => {
         allowRead: expect.arrayContaining([canonicalWorkspace, "/usr/bin"]),
       },
       credentials: {
-        envVars: expect.arrayContaining([
-          { name: "ANTHROPIC_AUTH_TOKEN", mode: "deny" },
-          { name: "ANTHROPIC_API_KEY", mode: "deny" },
-        ]),
+        envVars: [],
       },
     });
   });

hub/test/unit/provider-proxy.test.ts

@@ -53,13 +53,14 @@ describe("run-scoped provider proxy", () => {
     expect(lease.sdkEnv).toMatchObject({
       ANTHROPIC_BASE_URL: expect.stringMatching(/^http:\/\/127\.0\.0\.1:\d+$/),
       ANTHROPIC_AUTH_TOKEN: expect.any(String),
-      ANTHROPIC_API_KEY: "",
+      ANTHROPIC_API_KEY: expect.any(String),
     });
+    expect(lease.sdkEnv.ANTHROPIC_AUTH_TOKEN).toBe(lease.sdkEnv.ANTHROPIC_API_KEY);
 
     const response = await fetch(`${lease.sdkEnv.ANTHROPIC_BASE_URL}/v1/messages`, {
       method: "POST",
       headers: {
-        authorization: `Bearer ${lease.sdkEnv.ANTHROPIC_AUTH_TOKEN}`,
+        "x-api-key": lease.sdkEnv.ANTHROPIC_API_KEY ?? "",
         "content-type": "application/json",
         "anthropic-version": "2023-06-01",
       },
@@ -104,7 +105,38 @@ describe("run-scoped provider proxy", () => {
 
     expect(response.status).toBe(401);
     expect(upstreamRequests).toBe(0);
-    expect(diagnostics).toEqual([{ code: "provider_proxy_unauthorized", category: "authorization" }]);
+    expect(diagnostics).toEqual([{
+      code: "provider_proxy_unauthorized",
+      category: "authorization",
+      authShape: { bearerLength: 0, apiKeyLength: 0, expectedLength: 43 },
+    }]);
+  });
+
+  it("accepts the run capability in the Anthropic x-api-key header", async () => {
+    let upstreamRequests = 0;
+    const upstream = createServer((_request, response) => {
+      upstreamRequests += 1;
+      response.writeHead(200, { "content-type": "application/json" });
+      response.end(JSON.stringify({ ok: true }));
+    });
+    upstreamServers.push(upstream);
+    upstream.listen(0, "127.0.0.1");
+    await once(upstream, "listening");
+    const address = upstream.address();
+    if (address === null || typeof address === "string") throw new Error("expected upstream TCP address");
+    const lease = await openProviderProxyLease({
+      baseUrl: `http://127.0.0.1:${address.port}`,
+      authToken: "customer-secret",
+      anthropicApiKey: "",
+    });
+    leases.push(lease);
+
+    const response = await fetch(`${lease.sdkEnv.ANTHROPIC_BASE_URL}/v1/messages`, {
+      headers: { "x-api-key": lease.sdkEnv.ANTHROPIC_API_KEY ?? "" },
+    });
+
+    expect(response.status).toBe(200);
+    expect(upstreamRequests).toBe(1);
   });
 
   it("does not forward provider credentials across an upstream redirect", async () => {
@@ -140,7 +172,7 @@ describe("run-scoped provider proxy", () => {
     leases.push(lease);
 
     const response = await fetch(`${lease.sdkEnv.ANTHROPIC_BASE_URL}/v1/messages`, {
-      headers: { authorization: `Bearer ${lease.sdkEnv.ANTHROPIC_AUTH_TOKEN}` },
+      headers: { "x-api-key": lease.sdkEnv.ANTHROPIC_API_KEY ?? "" },
       redirect: "manual",
     });
 

hub/test/unit/runner.test.ts

@@ -288,10 +288,7 @@ describe("runAgent", () => {
         },
         sandbox: {
           credentials: {
-            envVars: expect.arrayContaining([
-              { name: "ANTHROPIC_AUTH_TOKEN", mode: "deny" },
-              { name: "ANTHROPIC_API_KEY", mode: "deny" },
-            ]),
+            envVars: [],
           },
         },
       },