diff --git a/gardener/best-practices.md b/gardener/best-practices.md
index d1ac8c1..cc75d79 100644
--- a/gardener/best-practices.md
+++ b/gardener/best-practices.md
@@ -21,7 +21,7 @@ A "ready" issue has:
 ## When to escalate (NEVER decide these)
 - Issue scope is ambiguous — could be interpreted multiple ways
 - Two issues overlap but aren't exact duplicates — need human to pick scope
-- Issue contradicts a design decision (check PRODUCT-TRUTH.md, ARCHITECTURE.md)
+- Issue contradicts a design decision (check AGENTS.md tree)
 - Issue is feature request vs bug — classification matters for priority
 - Closing would lose important context that isn't captured elsewhere
 
diff --git a/gardener/gardener-poll.sh b/gardener/gardener-poll.sh
index b65a714..ced23ff 100755
--- a/gardener/gardener-poll.sh
+++ b/gardener/gardener-poll.sh
@@ -182,7 +182,7 @@ $(echo -e "$PROBLEMS")
 ## Primary mission: promote tech-debt → backlog
 Most open issues are raw review-bot findings labeled \`tech-debt\`. Your main job is to convert them into well-structured \`backlog\` items the dev-agent can execute. For each tech-debt issue:
 1. Read the issue body + referenced source files to understand the real problem
-2. Check docs/PRODUCT-TRUTH.md and docs/ARCHITECTURE.md for context
+2. Check AGENTS.md (and sub-directory AGENTS.md files) for architecture context
 3. Add missing sections: \`## Affected files\`, \`## Acceptance criteria\` (checkboxes, max 5), \`## Dependencies\`
 4. If the issue is clear and actionable → relabel: remove \`tech-debt\`, add \`backlog\`
 5. If scope is ambiguous or needs a design decision → ESCALATE with options
diff --git a/review/review-pr.sh b/review/review-pr.sh
index b4d1399..324d8c5 100755
--- a/review/review-pr.sh
+++ b/review/review-pr.sh
@@ -301,7 +301,7 @@ You are running in a checkout of the PR branch. You can read ANY file in the rep
 claims, check existing code, or understand context. Use this to avoid false positives —
 if you're unsure whether something "already exists", read the file before flagging it.
 
-Key docs available: docs/PRODUCT-TRUTH.md, docs/ARCHITECTURE.md, docs/UX-DECISIONS.md, docs/ENVIRONMENT.md
+Key docs available: AGENTS.md (root + sub-directory files), docs/UX-DECISIONS.md, docs/ENVIRONMENT.md
 PROMPT_EOF
 
 if [ "$DIFF_TRUNCATED" = true ]; then
@@ -357,10 +357,10 @@ else
   # Build task description based on scope
   TASK_DESC="Review this ${SCOPE_DESC} PR."
   if [ "$NEEDS_CLAIM_CHECK" = true ]; then
-    TASK_DESC="${TASK_DESC} Check all user-facing claims against docs/PRODUCT-TRUTH.md."
+    TASK_DESC="${TASK_DESC} Check all user-facing claims against AGENTS.md."
   fi
   TASK_DESC="${TASK_DESC} Check for bugs, logic errors, missing edge cases, broken imports."
-  TASK_DESC="${TASK_DESC} Verify architecture patterns match docs/ARCHITECTURE.md."
+  TASK_DESC="${TASK_DESC} Verify architecture patterns match AGENTS.md."
   if [ "$NEEDS_UX_CHECK" = true ]; then
     TASK_DESC="${TASK_DESC} Check UX/messaging against docs/UX-DECISIONS.md."
   fi
@@ -616,7 +616,7 @@ COMMENT_BODY="## 🤖 AI ${REVIEW_TYPE}
 ${REVIEW_MD}
 
 ---
-*Reviewed at \`${PR_SHA:0:7}\`$(if [ "$IS_RE_REVIEW" = true ]; then echo " · Previous: \`${PREV_REVIEW_SHA:0:7}\`"; fi) · [PRODUCT-TRUTH.md](../docs/PRODUCT-TRUTH.md) · [ARCHITECTURE.md](../docs/ARCHITECTURE.md)*"
+*Reviewed at \`${PR_SHA:0:7}\`$(if [ "$IS_RE_REVIEW" = true ]; then echo " · Previous: \`${PREV_REVIEW_SHA:0:7}\`"; fi) · [AGENTS.md](AGENTS.md)*"
 
 printf '%s' "$COMMENT_BODY" > "${TMPDIR}/comment-body.txt"
 jq -Rs '{body: .}' < "${TMPDIR}/comment-body.txt" > "${TMPDIR}/comment.json"