diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
index 4924bb435..d54365349 100644
--- a/.github/workflows/deploy.yml
+++ b/.github/workflows/deploy.yml
@@ -584,6 +584,50 @@ jobs:
                   echo "::error::OAuth redirect smoke summary: non_rate_limited_attempts=$non_rate_limited_attempts, rate_limited_count=$rate_limited_count"
                   exit 1
 
+            - name: Wait for homelab deploy completion
+              if: steps.docker_wait.outputs.docker_rebuilt == 'true'
+              env:
+                  GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+              run: |
+                  set -euo pipefail
+                  if [ "${{ github.event_name }}" = "workflow_run" ]; then
+                    expected="${{ github.event.workflow_run.head_sha }}"
+                  else
+                    expected="${{ github.sha }}"
+                  fi
+
+                  echo "==> Waiting for homelab deploy to complete (SHA: $expected)..."
+                  status_was_seen=false
+
+                  for attempt in $(seq 1 12); do
+                    echo "Attempt ${attempt}/12..."
+                    deploy_state=$(gh api \
+                      "repos/${{ github.repository }}/commits/${expected}/statuses" \
+                      --jq '[.[] | select(.context == "homelab-deploy")] | first | .state // ""' \
+                      2>/dev/null || echo "")
+
+                    [ "$deploy_state" = "pending" ] && status_was_seen=true
+
+                    if [ "$deploy_state" = "success" ]; then
+                      echo "==> Homelab deploy completed successfully."
+                      exit 0
+                    fi
+                    if [ "$deploy_state" = "failure" ] || [ "$deploy_state" = "error" ]; then
+                      echo "::error::Homelab deploy reported failure after image deployment — check deploy logs on homelab"
+                      exit 1
+                    fi
+
+                    if [ "$status_was_seen" = "false" ] && [ "$attempt" -eq 6 ]; then
+                      echo "::warning::No homelab-deploy status after ~90s — GITHUB_DEPLOY_STATUS_TOKEN likely not configured on homelab. Cannot confirm full deploy completion."
+                      exit 0
+                    fi
+
+                    echo "Deploy status: ${deploy_state:-none}. Waiting 15s..."
+                    sleep 15
+                  done
+
+                  echo "::warning::Deploy completion timed out after 3 min — bot health gate may still be running. Check Discord notification for full outcome."
+
             - name: Notify failure
               if: failure()
               run: echo "::error::Homelab deployment failed — check logs above"
diff --git a/docs/decisions/2026-05-24-deploy-ci-deploy-completion-wait.md b/docs/decisions/2026-05-24-deploy-ci-deploy-completion-wait.md
new file mode 100644
index 000000000..de8a40ca8
--- /dev/null
+++ b/docs/decisions/2026-05-24-deploy-ci-deploy-completion-wait.md
@@ -0,0 +1,86 @@
+# Deploy CI: wait for homelab deploy completion on `docker_rebuilt=true` path
+
+Date: 2026-05-24
+
+## Context
+
+The Lucky CI deploy pipeline fires via `workflow_run` after Docker images are published. The
+webhook handler runs `deploy.sh` asynchronously (`nohup ... & disown`), so the webhook
+returns before `deploy.sh` completes.
+
+Gap B (PR #1046) added commit status posting to `deploy.sh`: it posts `pending` on sync,
+then `success` or `failure` to the `homelab-deploy` GitHub commit status context on exit.
+
+CI's "Validate deployed version" step has two paths:
+
+- `docker_rebuilt=false` (no image rebuild): already polls `homelab-deploy` commit status,
+  waits for `success`/`failure`, and exits accordingly.
+- `docker_rebuilt=true` (image rebuilt): polls `/api/health/version` for SHA match and exits
+  0 on match — but does not wait for `deploy.sh` to complete.
+
+The race window: `/api/health/version` can return the new SHA as soon as the backend
+container restarts. But `deploy.sh` continues executing (bot health poll, up to 90s). If
+the bot health gate fails after CI sees the SHA match, CI marks the deploy step green while
+`deploy.sh` exits non-zero and posts `failure` to `homelab-deploy`. Signals diverge.
+
+## Decision
+
+Add a "Wait for homelab deploy completion" step after all smoke checks, conditioned on
+`docker_rebuilt=true`. The step polls the `homelab-deploy` commit status using `gh api`
+for up to 3 minutes (12 × 15s). Outcomes:
+
+- `success` → CI passes.
+- `failure`/`error` → CI fails with an explicit error.
+- No status seen after 90s → warn and proceed (`GITHUB_DEPLOY_STATUS_TOKEN` not configured on
+  homelab; cannot confirm full completion — accepted degraded mode).
+- Timeout (3 min) → warn and proceed (bot health gate is warn-on-timeout by design; Discord
+  notification carries the authoritative outcome).
+
+The `docker_rebuilt=false` path is unaffected — it already has equivalent logic in "Validate
+deployed version".
+
+## Alternatives considered
+
+**Do nothing (accept + document)** — Bot health gate is warn-on-timeout by design; SHA match
+and smoke checks are strong signals. Discord notification tells the operator about bot health.
+Rejected: an operator reading "CI green" after a deploy with a divergent `homelab-deploy:
+failure` status has to cross-check two systems to understand deploy outcome. Confusion is
+real and bounded (~90s race window but repeatable).
+
+**Make webhook synchronous** — Remove `nohup ... & disown` from `deploy-wrapper.sh`; webhook
+blocks until `deploy.sh` completes. True exit code reaches CI. Rejected: adds 3–5 min to
+webhook response time; homelab nginx timeout risk; requires deploy-wrapper redesign.
+
+**Fail CI on missing `GITHUB_DEPLOY_STATUS_TOKEN`** — Hard-fail the step if no status appears
+after 90s. Rejected: inconsistent with the `docker_rebuilt=false` path, which also falls
+back gracefully when the token is absent. A missing token means the feature is disabled, not
+that the deploy failed.
+
+## Consequences
+
+**Positive:**
+
+- CI outcome on `docker_rebuilt=true` now reflects the full deploy cycle, including bot health
+  gate, not just SHA publication.
+- Consistent behavior between `docker_rebuilt=true` and `docker_rebuilt=false` paths: both
+  wait for the `homelab-deploy` commit status before the job completes.
+- Adds at most ~90s to CI runtime (time from SHA match to bot gateway healthy).
+
+**Negative:**
+
+- Requires `GITHUB_DEPLOY_STATUS_TOKEN` to be configured on homelab. If absent, step warns
+  and proceeds — the race condition remains in degraded mode.
+- If `deploy.sh` exit trap fails to call `post_deploy_status`, the step times out (warn).
+
+**Neutral:**
+
+- The bot health gate's warn-on-timeout design is preserved end-to-end: a Discord outage
+  causes `deploy.sh` to warn (not fail), so `homelab-deploy` still reaches `success`, and CI
+  follows.
+
+## Revisit when
+
+- `GITHUB_DEPLOY_STATUS_TOKEN` stability breaks on homelab → flip to Option 4 + enhanced
+  Discord alerting (post bot health outcome to channel explicitly).
+- Bot health gate is changed from warn-on-timeout to fail-on-timeout → reassess whether
+  CI timeout warning is still appropriate.