diff --git a/examples/dark-factory/coder/entrypoint.js b/examples/dark-factory/coder/entrypoint.js
index 79b421cf..d7e87bc0 100644
--- a/examples/dark-factory/coder/entrypoint.js
+++ b/examples/dark-factory/coder/entrypoint.js
@@ -256,11 +256,25 @@ function runCoder(repoDir) {
   // Inherit stdio so the coder CLI's own output + errors stream into the pod
   // logs (kubectl logs), instead of being swallowed by execFileSync's exception.
   const opts = { cwd: repoDir, env, stdio: "inherit", maxBuffer: 64 * 1024 * 1024 };
-  const prompt = `Implement the change described in ${WORKSPACE}/SPEC.md. Build and run unit tests until green. Commit your work.`;
+  // The coder also writes a concise, human-readable summary of WHAT it changed to
+  // artifacts/description.md — this becomes the "Changes" section of the PR body
+  // (in addition to the verification section). Keep it short: what changed + why,
+  // as reviewer-facing markdown bullets.
+  const descPath = `${WORKSPACE}/artifacts/description.md`;
+  const prompt =
+    `Implement the change described in ${WORKSPACE}/SPEC.md. Build and run unit tests until green. Commit your work. ` +
+    `Then write a concise description of the changes you made (what changed and why, as a few markdown bullet points, ` +
+    `reviewer-facing — no preamble) to ${descPath}.`;
   if (ENGINE === "kiro") {
     // Kiro CLI headless — the coder image carries the `kiro` binary; it reads the
     // same Bifrost/Bedrock env above. --headless drives it non-interactively.
+    // Append the description instruction so kiro produces the same artifact.
     console.log("[coder] engine=kiro (kiro run --headless)");
+    try {
+      fs.appendFileSync(`${WORKSPACE}/SPEC.md`,
+        `\n\n---\n\n## After implementing\n\nWrite a concise description of the changes you made ` +
+        `(what changed and why, a few reviewer-facing markdown bullets, no preamble) to ${descPath}.\n`);
+    } catch { /* non-fatal */ }
     return execFileSync("kiro", ["run", "--headless", "--spec", `${WORKSPACE}/SPEC.md`], opts);
   }
   console.log("[coder] engine=claude (claude -p)");
@@ -429,9 +443,17 @@ async function main() {
         ? "- ⏳ **Security review (AWS Security Agent):** _queued (DevOps cleared)…_"
         : "- ⬜ **Security review (AWS Security Agent):** _waiting on DevOps clearance_";
     }
+    // Coder-authored description of the changes (artifacts/description.md). Shown
+    // as a "Changes" section ahead of the verification block. Falls back to a
+    // neutral line if the coder didn't produce one, so the PR body is never empty.
+    const desc = readSecret(`${WORKSPACE}/artifacts/description.md`);
+    const changesSection = desc
+      ? ["### 📝 Changes", "", desc, ""]
+      : ["### 📝 Changes", "", "_Implemented per the linked issue; see the diff for details._", ""];
     const prBody = [
       `Closes #${ISSUE}.`,
       "",
+      ...changesSection,
       "<!-- dark-factory:status -->",
       "### 🏭 Dark Factory — verification",
       `- ✅ **Build + unit tests:** ${test.summary}`,
diff --git a/gitops/addons/charts/dark-factory/scripts/security-agent.sh b/gitops/addons/charts/dark-factory/scripts/security-agent.sh
index 8dc938bb..9601df42 100644
--- a/gitops/addons/charts/dark-factory/scripts/security-agent.sh
+++ b/gitops/addons/charts/dark-factory/scripts/security-agent.sh
@@ -95,27 +95,53 @@ JOBID="$(aws securityagent start-code-review-job --region "$AWS_REGION" \
 log "codeReviewJobId=${JOBID} — polling (timeout ${POLL_TIMEOUT}s)..."
 
 # ── 4. Poll to completion ────────────────────────────────────────────────────
+# The job's `status` field lags well behind the actual analysis: the AWS Security
+# Agent GitHub App posts its findings comment on the PR (e.g. "No issues identified")
+# minutes before batch-get-code-review-jobs flips to COMPLETED. Polling status alone
+# therefore blocks this (advisory) step for the full timeout even though the result
+# is already known. So we ALSO probe list-findings each iteration: once it returns a
+# well-formed result (findingsSummaries key present), the review has produced output
+# and we can proceed immediately — this is the early-exit that avoids the long wait.
 DEADLINE=$(( $(date +%s) + POLL_TIMEOUT ))
 STATUS="IN_PROGRESS"
+DONE=""
+: > "$WORK/findings.json"
 while [ "$(date +%s)" -lt "$DEADLINE" ]; do
   STATUS="$(aws securityagent batch-get-code-review-jobs --region "$AWS_REGION" \
     --agent-space-id "$AGENT_SPACE_ID" --code-review-job-ids "$JOBID" \
     --query 'codeReviewJobs[0].status' --output text 2>/dev/null || echo IN_PROGRESS)"
   log "job status=${STATUS}"
   case "$STATUS" in
-    COMPLETED|SUCCEEDED) break ;;
+    COMPLETED|SUCCEEDED) DONE="status"; break ;;
     FAILED|STOPPED|ERROR) log "review job ${STATUS}"; post_status "error" "security: review job ${STATUS}"; exit 0 ;;
   esac
+  # Early-exit: findings ready before status flips? (App bot already posted them.)
+  if aws securityagent list-findings --region "$AWS_REGION" \
+       --agent-space-id "$AGENT_SPACE_ID" --code-review-job-id "$JOBID" \
+       > "$WORK/findings.try.json" 2>/dev/null \
+     && python3 -c 'import json,sys; sys.exit(0 if "findingsSummaries" in json.load(open(sys.argv[1])) else 1)' "$WORK/findings.try.json" 2>/dev/null; then
+    mv "$WORK/findings.try.json" "$WORK/findings.json"
+    DONE="findings"; log "findings ready (status=${STATUS}) — proceeding without waiting for status flip"; break
+  fi
   sleep 20
 done
-if [ "$STATUS" != "COMPLETED" ] && [ "$STATUS" != "SUCCEEDED" ]; then
-  log "timed out waiting for review (last=${STATUS})"; post_status "error" "security: review timed out"; exit 0
+if [ -z "$DONE" ]; then
+  # Advisory step: the App bot posts the authoritative result on the PR regardless,
+  # so a slow job-status flip is NOT a failure. Post a neutral pending status (not
+  # error) so the PR check isn't a misleading red, and continue.
+  log "review still running past ${POLL_TIMEOUT}s (last status=${STATUS}) — see the AWS Security Agent bot comment on the PR for the authoritative result"
+  post_status "pending" "security: review still running — see AWS Security Agent PR comment"
+  exit 0
 fi
 
 # ── 5. Fetch findings, render report + verdict (python3, no node) ────────────
-aws securityagent list-findings --region "$AWS_REGION" \
-  --agent-space-id "$AGENT_SPACE_ID" --code-review-job-id "$JOBID" \
-  > "$WORK/findings.json" 2>/dev/null || echo '{"findingsSummaries":[]}' > "$WORK/findings.json"
+# Reuse the findings we already fetched during the early-exit probe; only re-fetch
+# if we broke on the status flip (findings not yet captured).
+if [ "$DONE" = "status" ] || [ ! -s "$WORK/findings.json" ]; then
+  aws securityagent list-findings --region "$AWS_REGION" \
+    --agent-space-id "$AGENT_SPACE_ID" --code-review-job-id "$JOBID" \
+    > "$WORK/findings.json" 2>/dev/null || echo '{"findingsSummaries":[]}' > "$WORK/findings.json"
+fi
 
 python3 - "$WORK/findings.json" "$BLOCK_LEVEL" "$WORK/report.md" > "$WORK/verdict.env" <<'PY'
 import json, sys