Failure notification on the daily sync and add persist-credentials: false and a permissions.

.github/workflows/add-submodules.yml

@@ -25,6 +25,9 @@ concurrency:
   group: add-submodules-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   add-submodules:
     runs-on: ubuntu-latest
@@ -33,6 +36,7 @@ jobs:
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:
           token: ${{ secrets.SYNC_TOKEN }}
+          persist-credentials: false
 
       - name: Add submodules
         shell: bash

.github/workflows/assets/notify.sh

@@ -0,0 +1,93 @@
+# shellcheck shell=bash
+# Slack failure notifications for sync-translation and heartbeat workflows.
+# Source env.sh before notify.sh when using workflow globals (e.g. HEARTBEAT_MAX_AGE_HOURS).
+
+# Return workflow run URL for the given run id (defaults to GITHUB_RUN_ID).
+workflow_run_url() {
+  local run_id="${1:-${GITHUB_RUN_ID:-}}"
+  local server="${GITHUB_SERVER_URL:-https://github.com}"
+  echo "${server}/${GITHUB_REPOSITORY}/actions/runs/${run_id}"
+}
+
+# Extract matrix lang from a sync-local job name, e.g. "sync-local (zh_Hans)" → zh_Hans.
+# Prints nothing for non-matrix job names (e.g. discover).
+extract_sync_local_lang_from_job_name() {
+  local job_name="$1"
+  local lang
+  lang="$(sed -n 's/.*sync-local[[:space:]]*(\([^)]*\)).*/\1/p' <<<"$job_name")"
+  if [[ -n "$lang" && "$lang" != "$job_name" ]]; then
+    echo "$lang"
+  fi
+}
+
+# Read gh api /actions/runs/.../jobs JSON from stdin; emit human-readable failed-job lines.
+format_failed_jobs_summary() {
+  jq -r '
+    (if type == "object" and has("jobs") then .jobs else . end)
+    | .[]
+    | select(.conclusion == "failure")
+    | select(.name | test("notify-failure") | not)
+    | .name as $name
+    | ($name | if test("sync-local \\(")
+        then (capture("sync-local \\((?<lang>[^)]+)\\)") | .lang // "")
+        else "" end) as $lang
+    | if ($lang | length) > 0 then "  • \($name) — lang=\($lang)"
+      else "  • \($name)" end
+  '
+}
+
+# Build a Slack incoming-webhook JSON payload. Args: title run_url summary_lines detail (optional).
+build_slack_failure_payload() {
+  local title="$1" run_url="$2" summary="$3" detail="${4:-}"
+  local text="${title}"$'\n'"Run: ${run_url}"
+  if [[ -n "$summary" ]]; then
+    text+=$'\n'"Failed jobs:"$'\n'"${summary}"
+  fi
+  if [[ -n "$detail" ]]; then
+    text+=$'\n'"${detail}"
+  fi
+  jq -n --arg text "$text" '{text: $text}'
+}
+
+# POST JSON payload to SLACK_WEBHOOK_URL.
+send_slack_notification() {
+  local payload="$1"
+  if [[ -z "${SLACK_WEBHOOK_URL:-}" ]]; then
+    echo "error: SLACK_WEBHOOK_URL secret is not set." >&2
+    return 1
+  fi
+  curl -fsS -X POST -H 'Content-Type: application/json' --data "$payload" "$SLACK_WEBHOOK_URL"
+}
+
+# Fetch failed jobs for a workflow run and format summary lines.
+collect_workflow_failed_jobs_summary() {
+  local run_id="${1:-${GITHUB_RUN_ID:-}}"
+  gh run view "$run_id" --repo "$GITHUB_REPOSITORY" --json jobs \
+    | format_failed_jobs_summary
+}
+
+# Notify Slack about sync-translation workflow failures (discover or sync-local).
+notify_sync_translation_failure() {
+  local run_id="${1:-${GITHUB_RUN_ID:-}}"
+  local run_url summary payload
+  run_url="$(workflow_run_url "$run_id")"
+  summary="$(collect_workflow_failed_jobs_summary "$run_id")"
+  payload="$(build_slack_failure_payload "Sync translation failed" "$run_url" "$summary" "")"
+  send_slack_notification "$payload"
+}
+
+# Notify Slack when the daily sync heartbeat is stale.
+notify_heartbeat_stale() {
+  local last_ts="$1"
+  local run_url workflow_url detail payload
+  local server="${GITHUB_SERVER_URL:-https://github.com}"
+  local last_desc="${last_ts:-none on record}"
+
+  run_url="$(workflow_run_url)"
+  workflow_url="${server}/${GITHUB_REPOSITORY}/actions/workflows/sync-translation.yml"
+  detail="Daily sync heartbeat: last successful scheduled run is stale (last=${last_desc}, threshold=${HEARTBEAT_MAX_AGE_HOURS}h). The scheduled sync may have stopped."
+  detail+=$'\n'"Workflow: ${workflow_url}"
+
+  payload="$(build_slack_failure_payload "Sync heartbeat alert" "$run_url" "" "$detail")"
+  send_slack_notification "$payload"
+}

.github/workflows/heartbeat.yml

@@ -6,6 +6,8 @@
 # alerts when the last *successful* scheduled sync is older than a threshold, so a
 # silently missed run is caught. It does not depend on sync-translation.yml
 # running, so it still fires once that workflow has stopped.
+#
+# Required secret: SLACK_WEBHOOK_URL — Slack incoming webhook for stale-heartbeat alerts.
 
 name: Sync heartbeat
 
@@ -31,13 +33,16 @@ jobs:
       - name: Check the last successful scheduled sync
         env:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}
         run: |
           set -euo pipefail
 
           # shellcheck source=assets/env.sh
           source "$GITHUB_WORKSPACE/.github/workflows/assets/env.sh"
           # shellcheck source=assets/lib.sh
           source "$GITHUB_WORKSPACE/.github/workflows/assets/lib.sh"
+          # shellcheck source=assets/notify.sh
+          source "$GITHUB_WORKSPACE/.github/workflows/assets/notify.sh"
 
           begin_phase "$PHASE_HEARTBEAT" "Check daily sync heartbeat"
           trap '[[ -n "$CURRENT_PHASE" ]] && end_phase' EXIT ERR
@@ -61,10 +66,9 @@ jobs:
           max_age_seconds=$(( HEARTBEAT_MAX_AGE_HOURS * 3600 ))
 
           if heartbeat_run_is_stale "$last_epoch" "$now_epoch" "$max_age_seconds"; then
-            # Interim alert: fail the job. Swap this for the shared notification
-            # channel once the failure-notification work provides one.
             last_desc="${last_ts:-none on record}"
             echo "::error::Daily sync heartbeat: last successful scheduled run is stale (last=${last_desc}, threshold=${HEARTBEAT_MAX_AGE_HOURS}h). The scheduled sync may have stopped." >&2
+            notify_heartbeat_stale "$last_ts"
             end_phase
             exit 1
           fi

.github/workflows/sync-translation.yml

@@ -11,8 +11,10 @@
 # Manual trigger: POST /repos/{owner}/{repo}/dispatches
 #   Body: {"event_type": "sync-translation"}
 #
-# Required secret: SYNC_TOKEN — PAT with repo push to this repository.
-#   Submodule remotes in .gitmodules point at per-library org (set when running add-submodules / start-translation with SUBMODULES_ORG).
+# Required secrets:
+#   SYNC_TOKEN — PAT with repo push to this repository.
+#     Submodule remotes in .gitmodules point at per-library org (set when running add-submodules / start-translation with SUBMODULES_ORG).
+#   SLACK_WEBHOOK_URL — Slack incoming webhook for failure alerts (notify-failure job).
 
 name: Sync translation
 
@@ -148,3 +150,31 @@ jobs:
           end_phase
 
           echo "Done." >&2
+
+  notify-failure:
+    needs: [discover, sync-local]
+    if: >-
+      always() &&
+      (needs.discover.result == 'failure' || needs.sync-local.result == 'failure')
+    runs-on: ubuntu-latest
+    permissions:
+      actions: read
+    steps:
+      - name: Checkout repository (master)
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          persist-credentials: false
+
+      - name: Notify Slack on failure
+        env:
+          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          set -euo pipefail
+
+          # shellcheck source=assets/env.sh
+          source "$GITHUB_WORKSPACE/.github/workflows/assets/env.sh"
+          # shellcheck source=assets/notify.sh
+          source "$GITHUB_WORKSPACE/.github/workflows/assets/notify.sh"
+
+          notify_sync_translation_failure

CHANGELOG.md

@@ -22,6 +22,9 @@ are a separate namespace — see [README](README.md#releases) and
 - Lint/CI schema checks via pinned `check-jsonschema` (metaschema + fixture
   validation); bats validates captured Weblate request bodies against the schema
   and asserts success response fields.
+- Slack failure notifications for `sync-translation` (via `notify-failure` job)
+  and stale-sync alerts for `heartbeat` (shared `notify.sh` + `SLACK_WEBHOOK_URL`
+  secret).
 
 ### Changed
 

README.md

@@ -203,6 +203,7 @@ documented below.
 | `SYNC_TOKEN`    | all workflows       | PAT with **`repo`** scope; **`add-submodules`** also needs permission to create org repositories when creating new mirrors. |
 | `WEBLATE_URL`   | `start-translation` | Base URL of the Weblate instance (the workflow appends **`WEBLATE_ENDPOINT_PATH`**).                                |
 | `WEBLATE_TOKEN` | `start-translation` | API token for that endpoint.                                                                                                |
+| `SLACK_WEBHOOK_URL` | `sync-translation`, `heartbeat` | Slack incoming webhook URL for failure and stale-heartbeat alerts. |
 
 ## Repository variables
 

docs/GETTING-STARTED.md

@@ -55,10 +55,11 @@ Configure these on the translations repository **before** dispatching any workfl
 | Secret   | `SYNC_TOKEN`     | [README § Required secrets](../README.md#required-secrets) |
 | Secret   | `WEBLATE_URL`    | [README § Required secrets](../README.md#required-secrets) |
 | Secret   | `WEBLATE_TOKEN`  | [README § Required secrets](../README.md#required-secrets) |
+| Secret   | `SLACK_WEBHOOK_URL` | [README § Required secrets](../README.md#required-secrets) — Slack incoming webhook for `sync-translation` failure and `heartbeat` stale-sync alerts |
 | Variable | `LANG_CODES`     | [README § Repository variables](../README.md#repository-variables) |
 | Variable | `SUBMODULES_ORG` | [README § Repository variables](../README.md#repository-variables) |
 
-`SYNC_TOKEN`, `WEBLATE_URL`, and `WEBLATE_TOKEN` are required secrets;
+`SYNC_TOKEN`, `WEBLATE_URL`, `WEBLATE_TOKEN`, and `SLACK_WEBHOOK_URL` are required secrets;
 `LANG_CODES` and optional `SUBMODULES_ORG` are repository variables. See the
 linked README sections for scope, format, and which workflows consume each value.
 

scripts/lint.sh

@@ -169,6 +169,7 @@ echo "lint: actionlint ${ACTIONLINT_VERSION} ($("$ACTIONLINT_BIN" -version | hea
 "$SHELLCHECK_BIN" -x \
   .github/workflows/assets/env.sh \
   .github/workflows/assets/lib.sh \
+  .github/workflows/assets/notify.sh \
   .github/workflows/assets/translation.sh \
   .github/workflows/assets/add_submodules.sh \
   .github/workflows/assets/submodule_ops.sh \

tests/helpers/common.bash

@@ -41,6 +41,13 @@ load_submodule_ops() {
   source "$ASSETS_DIR/submodule_ops.sh"
 }
 
+load_notify() {
+  load_env
+  export GITHUB_RUN_ID="${GITHUB_RUN_ID:-12345}"
+  # shellcheck source=/dev/null
+  source "$ASSETS_DIR/notify.sh"
+}
+
 # Run a function and capture its exit code (works under set -e in callers).
 run_fn() {
   local errexit_was_on=0

tests/helpers/http_mock.bash

@@ -144,4 +144,56 @@ restore_curl_stub() {
     rm -rf "$CURL_WRAPPER_DIR"
   fi
   unset CURL_WRAPPER_DIR REAL_CURL _CURL_ORIG_PATH MOCK_CURL_TIMEOUT MOCK_CURL_EXIT
+  unset MOCK_SLACK_REQUEST_LOG MOCK_SLACK_STATUS
+}
+
+install_slack_curl_stub() {
+  _CURL_ORIG_PATH="$PATH"
+  local wrapper_dir
+  wrapper_dir="$(mktemp -d)"
+  REAL_CURL="$(command -v curl)"
+  export REAL_CURL
+  MOCK_SLACK_REQUEST_LOG="$(mktemp)"
+  export MOCK_SLACK_REQUEST_LOG
+  MOCK_SLACK_STATUS="${MOCK_SLACK_STATUS:-200}"
+  export MOCK_SLACK_STATUS
+  cat >"$wrapper_dir/curl" <<'EOF'
+#!/usr/bin/env bash
+if [[ -n "${MOCK_CURL_EXIT:-}" ]]; then
+  echo "mock curl: simulated exit ${MOCK_CURL_EXIT}" >&2
+  exit "$MOCK_CURL_EXIT"
+fi
+if [[ "${MOCK_CURL_TIMEOUT:-}" == "1" ]]; then
+  echo "mock curl: simulated timeout" >&2
+  exit 28
+fi
+if [[ -n "${SLACK_WEBHOOK_URL:-}" && " $* " == *" ${SLACK_WEBHOOK_URL} "* ]]; then
+  body=""
+  prev=""
+  for arg in "$@"; do
+    if [[ "$prev" == "--data" || "$prev" == "-d" ]]; then
+      body="$arg"
+    fi
+    prev="$arg"
+  done
+  {
+    echo "URL=$SLACK_WEBHOOK_URL"
+    echo "BODY_START"
+    printf '%s' "$body"
+    echo
+    echo "BODY_END"
+  } >"${MOCK_SLACK_REQUEST_LOG}"
+  exit 0
+fi
+exec "$REAL_CURL" "$@"
+EOF
+  chmod +x "$wrapper_dir/curl"
+  CURL_WRAPPER_DIR="$wrapper_dir"
+  export PATH="$wrapper_dir:$PATH"
+}
+
+restore_slack_curl_stub() {
+  [[ -n "${MOCK_SLACK_REQUEST_LOG:-}" && -f "$MOCK_SLACK_REQUEST_LOG" ]] && rm -f "$MOCK_SLACK_REQUEST_LOG"
+  unset MOCK_SLACK_REQUEST_LOG MOCK_SLACK_STATUS
+  restore_curl_stub
 }