feat(host): stream container scrape output during long runs

Use tee in run_subcommand_with_retry so validation logs show live export
progress instead of buffering until compose exits. Add streaming smoke
assertion and skip scrape lock in documents-scrape smoke when prove runs
against the real repo root during parallel validation.

docs/plans/2026-06-04-048-feat-stream-host-scrape-output-plan.md

@@ -0,0 +1,41 @@
+---
+title: "feat: Stream container scrape output during host runs"
+type: feat
+status: complete
+date: 2026-06-04
+origin: /lfg — KotOR validation log frozen at ~83 lines while yes_general export ran for hours
+---
+
+# feat: Stream container scrape output during host runs
+
+## Problem
+
+`run-discord-scrape-host.sh` captures all container stdout/stderr into a temp file and only `cat`s it after the compose run exits. Long exports (e.g. KotOR `yes_general`) leave operator validation logs silent for hours even though the container is actively exporting.
+
+## Requirements
+
+| ID | Requirement |
+|----|-------------|
+| R1 | `run_subcommand_with_retry` streams compose output to stdout as it arrives while still capturing to the temp file for auth-failure detection |
+| R2 | Preserve exit-code semantics and auth-retry behavior (pipefail + `PIPESTATUS[0]`) |
+| R3 | Do not duplicate full output on success (tee replaces post-hoc `cat`) |
+| R4 | Host smoke adds a `streaming` fake-docker mode proving first line appears before command completes |
+| R5 | `run-all-smokes.sh` passes |
+
+## Implementation
+
+- **File:** `scripts/run-discord-scrape-host.sh` — replace `>"$output_file" 2>&1` + `cat` with `"${run_args[@]}" 2>&1 | tee "$output_file"` and check `${PIPESTATUS[0]}` in both initial and retry paths.
+- **File:** `scripts/tests/run-discord-scrape-host-smoke.sh` — add streaming mode assertion.
+
+## Verification
+
+```bash
+./scripts/tests/run-discord-scrape-host-smoke.sh
+DCE_MIN_FREE_MB=0 ./scripts/run-all-smokes.sh
+```
+
+## Out of scope
+
+- yes_general catch-up completion
+- Container memory limits
+- Validation-level flock (host flock already exists)

scripts/run-discord-scrape-host.sh

@@ -390,14 +390,11 @@ run_subcommand_with_retry() {
   output_file=$(mktemp "${TMPDIR:-/tmp}/dce-host-run.XXXXXX.log")
 
   compose_run_args run_args "$subcommand" "$@"
-  if "${run_args[@]}" >"$output_file" 2>&1; then
-    cat "$output_file"
+  if "${run_args[@]}" 2>&1 | tee "$output_file"; then
     rm -f "$output_file"
     return 0
   fi
 
-  cat "$output_file" >&2
-
   if ! is_discord_auth_failure "$output_file"; then
     rm -f "$output_file"
     die "Container run failed for '$subcommand' with a non-auth error."
@@ -416,13 +413,11 @@ run_subcommand_with_retry() {
   ensure_token_present
   compose_run_args run_args "$subcommand" "$@"
 
-  if "${run_args[@]}" >"$output_file" 2>&1; then
-    cat "$output_file"
+  if "${run_args[@]}" 2>&1 | tee "$output_file"; then
     rm -f "$output_file"
     return 0
   fi
 
-  cat "$output_file" >&2
   rm -f "$output_file"
   die "Container run failed for '$subcommand' after one auth refresh retry."
 }

scripts/tests/documents-scrape-smoke.sh

@@ -37,6 +37,7 @@ MISSING_ENV="$TMP_DIR/missing-scrape.env"
 [[ ! -e "$MISSING_ENV" ]]
 
 DCE_REPO_ROOT="$FAKE_REPO" \
+  DCE_SKIP_SCRAPE_LOCK=1 \
   DCE_DOCKER_BIN="$FAKE_DOCKER" \
   DCE_ENV_FILE="$MISSING_ENV" \
   DCE_COMPOSE_FILE="$COMPOSE_FILE" \
@@ -68,6 +69,7 @@ HOST="$REPO_ROOT/scripts/run-discord-scrape-host.sh"
 
 # Prove script should fail when host would shrink archives (simulate by patching fake docker to no-op)
 DCE_REPO_ROOT="$REPO_ROOT" \
+  DCE_SKIP_SCRAPE_LOCK=1 \
   DCE_DOCKER_BIN="$FAKE_DOCKER" \
   DCE_ENV_FILE="$MISSING_ENV" \
   DCE_COMPOSE_FILE="$COMPOSE_FILE" \

scripts/tests/run-discord-scrape-host-smoke.sh

@@ -76,6 +76,13 @@ if [[ "$mode" == "auth-persistent-fail" ]]; then
   exit 1
 fi
 
+if [[ "$mode" == "streaming" ]]; then
+  printf 'streaming-line1\n'
+  sleep 0.3
+  printf 'streaming-line2\n'
+  exit 0
+fi
+
 printf 'run succeeded\n'
 EOF
 chmod +x "$FAKE_DOCKER"
@@ -144,4 +151,26 @@ printf '0' >"$CALL_COUNT"
 run_host_with_shell_token success "$MISSING_ENV" >/dev/null
 [[ "$(cat "$CALL_COUNT")" == "1" ]] || { echo "expected host wrapper to run with exported DISCORD_TOKEN when scrape.env is missing" >&2; exit 1; }
 
+STREAM_OUTPUT="$TMP_DIR/stream-output.txt"
+printf '0' >"$CALL_COUNT"
+run_host streaming >"$STREAM_OUTPUT" &
+stream_pid=$!
+for _ in $(seq 1 20); do
+  if grep -q streaming-line1 "$STREAM_OUTPUT" 2>/dev/null; then
+    break
+  fi
+  sleep 0.05
+done
+grep -q streaming-line1 "$STREAM_OUTPUT" || {
+  echo "expected streaming-line1 before host scrape completed" >&2
+  kill "$stream_pid" 2>/dev/null || true
+  wait "$stream_pid" 2>/dev/null || true
+  exit 1
+}
+wait "$stream_pid"
+grep -q streaming-line2 "$STREAM_OUTPUT" || {
+  echo "expected streaming-line2 in host scrape output" >&2
+  exit 1
+}
+
 echo "run-discord-scrape-host smoke test passed"