diff --git a/docs/IMPLEMENT-PLAN.md b/docs/IMPLEMENT-PLAN.md
index 2a0cb31..1ed7f12 100644
--- a/docs/IMPLEMENT-PLAN.md
+++ b/docs/IMPLEMENT-PLAN.md
@@ -132,6 +132,8 @@ Verify Superpowers execution dependencies exist in your agent skills root:
 - Executes milestones one-by-one, tracking stories in `story-tracker.md`.
 - Runs lint/typecheck/tests as a gate before each milestone review.
 - Sends each milestone to a reviewer CLI for approval (max rounds configurable, default 10).
+- Runs reviewer commands through `reviewer-runtime/run-review.sh` when available, with fallback to direct synchronous execution only if the helper is missing.
+- Captures reviewer stderr and helper status logs for diagnostics and retains them on failed, empty-output, or operator-decision review rounds.
 - Commits each milestone locally only after reviewer approval (does not push).
 - After all milestones approved, merges worktree branch to parent and deletes worktree.
 - Supports resume: detects existing worktree and `in-dev`/`completed` stories.
@@ -141,11 +143,38 @@ Verify Superpowers execution dependencies exist in your agent skills root:
 After each milestone is implemented and verified, the skill sends it to a second model for review:
 
 1. **Configure** — user picks a reviewer CLI (`codex`, `claude`, `cursor`) and model, or skips
-2. **Submit** — milestone spec, acceptance criteria, git diff, and verification output are written to a temp file and sent to the reviewer in read-only/ask mode
-3. **Feedback** — reviewer evaluates correctness, acceptance criteria, code quality, test coverage, security
-4. **Revise** — the implementing agent addresses each issue, re-verifies, and re-submits
-5. **Repeat** — up to max rounds (default 10) until the reviewer returns `VERDICT: APPROVED`
-6. **Approve** — milestone is marked approved in `story-tracker.md`
+2. **Prepare** — milestone payload and a bash reviewer command script are written to temp files
+3. **Run** — the command script is executed through `reviewer-runtime/run-review.sh` when installed
+4. **Feedback** — reviewer evaluates correctness, acceptance criteria, code quality, test coverage, security
+5. **Revise** — the implementing agent addresses each issue, re-verifies, and re-submits
+6. **Repeat** — up to max rounds (default 10) until the reviewer returns `VERDICT: APPROVED`
+7. **Approve** — milestone is marked approved in `story-tracker.md`
+
+### Runtime Artifacts
+
+The milestone review flow may create these temp artifacts:
+
+- `/tmp/milestone-<id>.md` — milestone review payload
+- `/tmp/milestone-review-<id>.md` — normalized review text
+- `/tmp/milestone-review-<id>.json` — raw Cursor JSON output
+- `/tmp/milestone-review-<id>.stderr` — reviewer stderr
+- `/tmp/milestone-review-<id>.status` — helper heartbeat/status log
+- `/tmp/milestone-review-<id>.runner.out` — helper-managed stdout from the reviewer command process
+- `/tmp/milestone-review-<id>.sh` — reviewer command script
+
+Status log lines use this format:
+
+```text
+ts=<ISO-8601> level=<info|warn|error> state=<running-silent|running-active|stall-warning|completed|completed-empty-output|failed|needs-operator-decision> elapsed_s=<int> pid=<int> stdout_bytes=<int> stderr_bytes=<int> note="<short message>"
+```
+
+`stall-warning` is a heartbeat/status-log state only. It is not a terminal review result.
+
+### Failure Handling
+
+- `completed-empty-output` means the reviewer exited without producing review text; surface `.stderr` and `.status`, then retry only after diagnosing the cause.
+- `needs-operator-decision` means the helper reached hard-timeout escalation; surface `.status` and decide whether to keep waiting, abort, or retry with different parameters.
+- Successful rounds clean up temp artifacts. Failed, empty-output, and operator-decision rounds should retain `.stderr`, `.status`, and `.runner.out` until diagnosed.
 
 ### Supported Reviewer CLIs
 
@@ -155,6 +184,26 @@ After each milestone is implemented and verified, the skill sends it to a second
 | `claude` | `claude -p --model <model> --allowedTools Read` | No (fresh call each round) | `--allowedTools Read` |
 | `cursor` | `cursor-agent -p --mode=ask --model <model> --trust --output-format json` | Yes (`--resume <id>`) | `--mode=ask` |
 
+For all three CLIs, the preferred execution path is:
+
+1. write the reviewer command to a bash script
+2. run that script through `reviewer-runtime/run-review.sh`
+3. fall back to direct synchronous execution only if the helper is missing or not executable
+
+The helper also supports manual override flags for diagnostics:
+
+```bash
+run-review.sh \
+  --command-file <path> \
+  --stdout-file <path> \
+  --stderr-file <path> \
+  --status-file <path> \
+  --poll-seconds 10 \
+  --soft-timeout-seconds 600 \
+  --stall-warning-seconds 300 \
+  --hard-timeout-seconds 1800
+```
+
 ## Variant Hardening Notes
 
 ### Claude Code
diff --git a/skills/implement-plan/claude-code/SKILL.md b/skills/implement-plan/claude-code/SKILL.md
index e739dc8..1f702b1 100644
--- a/skills/implement-plan/claude-code/SKILL.md
+++ b/skills/implement-plan/claude-code/SKILL.md
@@ -136,7 +136,20 @@ Do NOT push. After committing:
 REVIEW_ID=$(uuidgen | tr '[:upper:]' '[:lower:]' | head -c 8)
 ```
 
-Use for all temp file paths: `/tmp/milestone-${REVIEW_ID}.md` and `/tmp/milestone-review-${REVIEW_ID}.md`.
+Use `REVIEW_ID` for all milestone review temp file paths:
+- `/tmp/milestone-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.json`
+- `/tmp/milestone-review-${REVIEW_ID}.stderr`
+- `/tmp/milestone-review-${REVIEW_ID}.status`
+- `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+- `/tmp/milestone-review-${REVIEW_ID}.sh`
+
+Resolve the shared runtime helper path before writing the command script:
+
+```bash
+REVIEWER_RUNTIME=~/.claude/skills/reviewer-runtime/run-review.sh
+```
 
 #### Step 2: Write Review Payload
 
@@ -165,6 +178,13 @@ Write to `/tmp/milestone-${REVIEW_ID}.md`:
 
 #### Step 3: Submit to Reviewer (Round 1)
 
+Write the reviewer invocation to `/tmp/milestone-review-${REVIEW_ID}.sh` as a bash script:
+
+```bash
+#!/usr/bin/env bash
+set -euo pipefail
+```
+
 **If `REVIEWER_CLI` is `codex`:**
 
 ```bash
@@ -185,7 +205,7 @@ Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE"
 ```
 
-Capture the Codex session ID from output (line `session id: <uuid>`). Store as `CODEX_SESSION_ID`.
+Do not try to capture the Codex session ID yet. When using the helper, extract it from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the command completes (look for `session id: <uuid>`), then store it as `CODEX_SESSION_ID` for resume in subsequent rounds.
 
 **If `REVIEWER_CLI` is `claude`:**
 
@@ -203,8 +223,7 @@ Evaluate:
 Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE" \
   --model ${REVIEWER_MODEL} \
-  --allowedTools Read \
-  > /tmp/milestone-review-${REVIEW_ID}.md
+  --allowedTools Read
 ```
 
 **If `REVIEWER_CLI` is `cursor`:**
@@ -229,19 +248,43 @@ If changes are needed, end with exactly: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
 ```
 
-Extract session ID and review text (requires `jq`):
+For `cursor`, the command script writes raw JSON to `/tmp/milestone-review-${REVIEW_ID}.json`. Do not run `jq` extraction until after the helper or fallback execution completes. If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivalent.
+
+Run the command script through the shared helper when available:
+
+```bash
+if [ -x "$REVIEWER_RUNTIME" ]; then
+  "$REVIEWER_RUNTIME" \
+    --command-file /tmp/milestone-review-${REVIEW_ID}.sh \
+    --stdout-file /tmp/milestone-review-${REVIEW_ID}.runner.out \
+    --stderr-file /tmp/milestone-review-${REVIEW_ID}.stderr \
+    --status-file /tmp/milestone-review-${REVIEW_ID}.status
+else
+  echo "Warning: reviewer runtime helper not found at $REVIEWER_RUNTIME; falling back to direct synchronous review." >&2
+  bash /tmp/milestone-review-${REVIEW_ID}.sh >/tmp/milestone-review-${REVIEW_ID}.runner.out 2>/tmp/milestone-review-${REVIEW_ID}.stderr
+fi
+```
+
+After the command completes:
+- If `REVIEWER_CLI=cursor`, extract the final review text:
 
 ```bash
 CURSOR_SESSION_ID=$(jq -r '.session_id' /tmp/milestone-review-${REVIEW_ID}.json)
 jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
-If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivalent.
+- If `REVIEWER_CLI=codex`, extract `CODEX_SESSION_ID` from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the helper or fallback run. If the review text is only in `.runner.out`, move or copy the actual review body into `/tmp/milestone-review-${REVIEW_ID}.md` before verdict parsing.
+
+Fallback is allowed only when the helper is missing or not executable.
 
 #### Step 4: Read Review & Check Verdict
 
 1. Read `/tmp/milestone-review-${REVIEW_ID}.md`
-2. Present review to the user:
+2. If the review failed, produced empty output, or reached helper timeout, also read:
+   - `/tmp/milestone-review-${REVIEW_ID}.stderr`
+   - `/tmp/milestone-review-${REVIEW_ID}.status`
+   - `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+3. Present review to the user:
 
 ```
 ## Milestone Review — Round N (reviewer: ${REVIEWER_CLI} / ${REVIEWER_MODEL})
@@ -249,10 +292,12 @@ If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivale
 [Reviewer feedback]
 ```
 
-3. Check verdict:
+4. Check verdict:
    - **VERDICT: APPROVED** -> proceed to Phase 4 Step 6 (commit & approve)
    - **VERDICT: REVISE** -> go to Step 5
    - No clear verdict but positive / no actionable items -> treat as approved
+   - Helper state `completed-empty-output` -> treat as failed review attempt, surface stderr/status, fix invocation or prompt handling, then retry
+   - Helper state `needs-operator-decision` -> surface status log, note any `stall-warning` heartbeat lines as non-terminal operator hints, and decide whether to keep waiting, abort, or retry with different helper parameters
    - Max rounds (`MAX_ROUNDS`) reached -> present to user for manual decision (proceed or stop)
 
 #### Step 5: Address Feedback & Re-verify
@@ -272,6 +317,8 @@ If a revision contradicts the user's explicit requirements, skip it and note it
 
 #### Step 6: Re-submit to Reviewer (Rounds 2-N)
 
+Rewrite `/tmp/milestone-review-${REVIEW_ID}.sh` for the next round. The script should contain the reviewer invocation only; do not run it directly.
+
 **If `REVIEWER_CLI` is `codex`:**
 
 Resume the existing session:
@@ -330,20 +377,31 @@ Changes made:
 Re-review. If solid, end with: VERDICT: APPROVED
 If more changes needed, end with: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
-
-jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
 If resume fails, fall back to fresh `cursor-agent -p` with context about prior rounds.
 
+Do not run `jq` extraction until after the helper or fallback execution completes, then extract `/tmp/milestone-review-${REVIEW_ID}.md` from the JSON response.
+
+After updating `/tmp/milestone-review-${REVIEW_ID}.sh`, run the same helper/fallback flow from Round 1.
+
 Return to Step 4.
 
 #### Step 7: Cleanup Per Milestone
 
 ```bash
-rm -f /tmp/milestone-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.json
+rm -f \
+  /tmp/milestone-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.json \
+  /tmp/milestone-review-${REVIEW_ID}.stderr \
+  /tmp/milestone-review-${REVIEW_ID}.status \
+  /tmp/milestone-review-${REVIEW_ID}.runner.out \
+  /tmp/milestone-review-${REVIEW_ID}.sh
 ```
 
+If the round failed, produced empty output, or reached operator-decision timeout, keep `.stderr`, `.status`, and `.runner.out` until the issue is diagnosed instead of deleting them immediately.
+
 ### Phase 6: Completion (REQUIRED SUB-SKILL)
 
 After all milestones are approved and committed:
diff --git a/skills/implement-plan/codex/SKILL.md b/skills/implement-plan/codex/SKILL.md
index e16b3be..f748376 100644
--- a/skills/implement-plan/codex/SKILL.md
+++ b/skills/implement-plan/codex/SKILL.md
@@ -169,7 +169,20 @@ Do NOT push. After committing:
 REVIEW_ID=$(uuidgen | tr '[:upper:]' '[:lower:]' | head -c 8)
 ```
 
-Use for all temp file paths: `/tmp/milestone-${REVIEW_ID}.md` and `/tmp/milestone-review-${REVIEW_ID}.md`.
+Use `REVIEW_ID` for all milestone review temp file paths:
+- `/tmp/milestone-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.json`
+- `/tmp/milestone-review-${REVIEW_ID}.stderr`
+- `/tmp/milestone-review-${REVIEW_ID}.status`
+- `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+- `/tmp/milestone-review-${REVIEW_ID}.sh`
+
+Resolve the shared runtime helper path before writing the command script:
+
+```bash
+REVIEWER_RUNTIME=~/.codex/skills/reviewer-runtime/run-review.sh
+```
 
 #### Step 2: Write Review Payload
 
@@ -198,6 +211,13 @@ Write to `/tmp/milestone-${REVIEW_ID}.md`:
 
 #### Step 3: Submit to Reviewer (Round 1)
 
+Write the reviewer invocation to `/tmp/milestone-review-${REVIEW_ID}.sh` as a bash script:
+
+```bash
+#!/usr/bin/env bash
+set -euo pipefail
+```
+
 **If `REVIEWER_CLI` is `codex`:**
 
 ```bash
@@ -218,7 +238,7 @@ Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE"
 ```
 
-Capture the Codex session ID from output (line `session id: <uuid>`). Store as `CODEX_SESSION_ID`.
+Do not try to capture the Codex session ID yet. When using the helper, extract it from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the command completes (look for `session id: <uuid>`), then store it as `CODEX_SESSION_ID` for resume in subsequent rounds.
 
 **If `REVIEWER_CLI` is `claude`:**
 
@@ -236,8 +256,7 @@ Evaluate:
 Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE" \
   --model ${REVIEWER_MODEL} \
-  --allowedTools Read \
-  > /tmp/milestone-review-${REVIEW_ID}.md
+  --allowedTools Read
 ```
 
 **If `REVIEWER_CLI` is `cursor`:**
@@ -262,19 +281,43 @@ If changes are needed, end with exactly: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
 ```
 
-Extract session ID and review text (requires `jq`):
+For `cursor`, the command script writes raw JSON to `/tmp/milestone-review-${REVIEW_ID}.json`. Do not run `jq` extraction until after the helper or fallback execution completes. If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivalent.
+
+Run the command script through the shared helper when available:
+
+```bash
+if [ -x "$REVIEWER_RUNTIME" ]; then
+  "$REVIEWER_RUNTIME" \
+    --command-file /tmp/milestone-review-${REVIEW_ID}.sh \
+    --stdout-file /tmp/milestone-review-${REVIEW_ID}.runner.out \
+    --stderr-file /tmp/milestone-review-${REVIEW_ID}.stderr \
+    --status-file /tmp/milestone-review-${REVIEW_ID}.status
+else
+  echo "Warning: reviewer runtime helper not found at $REVIEWER_RUNTIME; falling back to direct synchronous review." >&2
+  bash /tmp/milestone-review-${REVIEW_ID}.sh >/tmp/milestone-review-${REVIEW_ID}.runner.out 2>/tmp/milestone-review-${REVIEW_ID}.stderr
+fi
+```
+
+After the command completes:
+- If `REVIEWER_CLI=cursor`, extract the final review text:
 
 ```bash
 CURSOR_SESSION_ID=$(jq -r '.session_id' /tmp/milestone-review-${REVIEW_ID}.json)
 jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
-If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivalent.
+- If `REVIEWER_CLI=codex`, extract `CODEX_SESSION_ID` from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the helper or fallback run. If the review text is only in `.runner.out`, move or copy the actual review body into `/tmp/milestone-review-${REVIEW_ID}.md` before verdict parsing.
+
+Fallback is allowed only when the helper is missing or not executable.
 
 #### Step 4: Read Review & Check Verdict
 
 1. Read `/tmp/milestone-review-${REVIEW_ID}.md`
-2. Present review to the user:
+2. If the review failed, produced empty output, or reached helper timeout, also read:
+   - `/tmp/milestone-review-${REVIEW_ID}.stderr`
+   - `/tmp/milestone-review-${REVIEW_ID}.status`
+   - `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+3. Present review to the user:
 
 ```
 ## Milestone Review — Round N (reviewer: ${REVIEWER_CLI} / ${REVIEWER_MODEL})
@@ -282,10 +325,12 @@ If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivale
 [Reviewer feedback]
 ```
 
-3. Check verdict:
+4. Check verdict:
    - **VERDICT: APPROVED** -> proceed to Phase 4 Step 6 (commit & approve)
    - **VERDICT: REVISE** -> go to Step 5
    - No clear verdict but positive / no actionable items -> treat as approved
+   - Helper state `completed-empty-output` -> treat as failed review attempt, surface stderr/status, fix invocation or prompt handling, then retry
+   - Helper state `needs-operator-decision` -> surface status log, note any `stall-warning` heartbeat lines as non-terminal operator hints, and decide whether to keep waiting, abort, or retry with different helper parameters
    - Max rounds (`MAX_ROUNDS`) reached -> present to user for manual decision (proceed or stop)
 
 #### Step 5: Address Feedback & Re-verify
@@ -305,6 +350,8 @@ If a revision contradicts the user's explicit requirements, skip it and note it
 
 #### Step 6: Re-submit to Reviewer (Rounds 2-N)
 
+Rewrite `/tmp/milestone-review-${REVIEW_ID}.sh` for the next round. The script should contain the reviewer invocation only; do not run it directly.
+
 **If `REVIEWER_CLI` is `codex`:**
 
 Resume the existing session:
@@ -363,20 +410,31 @@ Changes made:
 Re-review. If solid, end with: VERDICT: APPROVED
 If more changes needed, end with: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
-
-jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
 If resume fails, fall back to fresh `cursor-agent -p` with context about prior rounds.
 
+Do not run `jq` extraction until after the helper or fallback execution completes, then extract `/tmp/milestone-review-${REVIEW_ID}.md` from the JSON response.
+
+After updating `/tmp/milestone-review-${REVIEW_ID}.sh`, run the same helper/fallback flow from Round 1.
+
 Return to Step 4.
 
 #### Step 7: Cleanup Per Milestone
 
 ```bash
-rm -f /tmp/milestone-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.json
+rm -f \
+  /tmp/milestone-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.json \
+  /tmp/milestone-review-${REVIEW_ID}.stderr \
+  /tmp/milestone-review-${REVIEW_ID}.status \
+  /tmp/milestone-review-${REVIEW_ID}.runner.out \
+  /tmp/milestone-review-${REVIEW_ID}.sh
 ```
 
+If the round failed, produced empty output, or reached operator-decision timeout, keep `.stderr`, `.status`, and `.runner.out` until the issue is diagnosed instead of deleting them immediately.
+
 ### Phase 6: Completion (REQUIRED SUB-SKILL)
 
 After all milestones are approved and committed:
diff --git a/skills/implement-plan/cursor/SKILL.md b/skills/implement-plan/cursor/SKILL.md
index 9191171..17f4846 100644
--- a/skills/implement-plan/cursor/SKILL.md
+++ b/skills/implement-plan/cursor/SKILL.md
@@ -169,7 +169,24 @@ Do NOT push. After committing:
 REVIEW_ID=$(uuidgen | tr '[:upper:]' '[:lower:]' | head -c 8)
 ```
 
-Use for all temp file paths: `/tmp/milestone-${REVIEW_ID}.md` and `/tmp/milestone-review-${REVIEW_ID}.md`.
+Use `REVIEW_ID` for all milestone review temp file paths:
+- `/tmp/milestone-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.json`
+- `/tmp/milestone-review-${REVIEW_ID}.stderr`
+- `/tmp/milestone-review-${REVIEW_ID}.status`
+- `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+- `/tmp/milestone-review-${REVIEW_ID}.sh`
+
+Resolve the shared runtime helper path before writing the command script:
+
+```bash
+if [ -x .cursor/skills/reviewer-runtime/run-review.sh ]; then
+  REVIEWER_RUNTIME=.cursor/skills/reviewer-runtime/run-review.sh
+else
+  REVIEWER_RUNTIME=~/.cursor/skills/reviewer-runtime/run-review.sh
+fi
+```
 
 #### Step 2: Write Review Payload
 
@@ -198,6 +215,13 @@ Write to `/tmp/milestone-${REVIEW_ID}.md`:
 
 #### Step 3: Submit to Reviewer (Round 1)
 
+Write the reviewer invocation to `/tmp/milestone-review-${REVIEW_ID}.sh` as a bash script:
+
+```bash
+#!/usr/bin/env bash
+set -euo pipefail
+```
+
 **If `REVIEWER_CLI` is `codex`:**
 
 ```bash
@@ -218,7 +242,7 @@ Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE"
 ```
 
-Capture the Codex session ID from output (line `session id: <uuid>`). Store as `CODEX_SESSION_ID`.
+Do not try to capture the Codex session ID yet. When using the helper, extract it from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the command completes (look for `session id: <uuid>`), then store it as `CODEX_SESSION_ID` for resume in subsequent rounds.
 
 **If `REVIEWER_CLI` is `claude`:**
 
@@ -236,8 +260,7 @@ Evaluate:
 Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE" \
   --model ${REVIEWER_MODEL} \
-  --allowedTools Read \
-  > /tmp/milestone-review-${REVIEW_ID}.md
+  --allowedTools Read
 ```
 
 **If `REVIEWER_CLI` is `cursor`:**
@@ -262,12 +285,7 @@ If changes are needed, end with exactly: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
 ```
 
-Extract session ID and review text:
-
-```bash
-CURSOR_SESSION_ID=$(jq -r '.session_id' /tmp/milestone-review-${REVIEW_ID}.json)
-jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
-```
+For `cursor`, the command script writes raw JSON to `/tmp/milestone-review-${REVIEW_ID}.json`. Do not run `jq` extraction until after the helper or fallback execution completes.
 
 Notes on Cursor flags:
 - `--mode=ask` — read-only mode, no file modifications
@@ -275,10 +293,41 @@ Notes on Cursor flags:
 - `-p` / `--print` — non-interactive mode, output to stdout
 - `--output-format json` — structured output with `session_id` and `result` fields
 
+Run the command script through the shared helper when available:
+
+```bash
+if [ -x "$REVIEWER_RUNTIME" ]; then
+  "$REVIEWER_RUNTIME" \
+    --command-file /tmp/milestone-review-${REVIEW_ID}.sh \
+    --stdout-file /tmp/milestone-review-${REVIEW_ID}.runner.out \
+    --stderr-file /tmp/milestone-review-${REVIEW_ID}.stderr \
+    --status-file /tmp/milestone-review-${REVIEW_ID}.status
+else
+  echo "Warning: reviewer runtime helper not found at $REVIEWER_RUNTIME; falling back to direct synchronous review." >&2
+  bash /tmp/milestone-review-${REVIEW_ID}.sh >/tmp/milestone-review-${REVIEW_ID}.runner.out 2>/tmp/milestone-review-${REVIEW_ID}.stderr
+fi
+```
+
+After the command completes:
+- If `REVIEWER_CLI=cursor`, extract the final review text:
+
+```bash
+CURSOR_SESSION_ID=$(jq -r '.session_id' /tmp/milestone-review-${REVIEW_ID}.json)
+jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
+```
+
+- If `REVIEWER_CLI=codex`, extract `CODEX_SESSION_ID` from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the helper or fallback run. If the review text is only in `.runner.out`, move or copy the actual review body into `/tmp/milestone-review-${REVIEW_ID}.md` before verdict parsing.
+
+Fallback is allowed only when the helper is missing or not executable.
+
 #### Step 4: Read Review & Check Verdict
 
 1. Read `/tmp/milestone-review-${REVIEW_ID}.md`
-2. Present review to the user:
+2. If the review failed, produced empty output, or reached helper timeout, also read:
+   - `/tmp/milestone-review-${REVIEW_ID}.stderr`
+   - `/tmp/milestone-review-${REVIEW_ID}.status`
+   - `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+3. Present review to the user:
 
 ```
 ## Milestone Review — Round N (reviewer: ${REVIEWER_CLI} / ${REVIEWER_MODEL})
@@ -286,10 +335,12 @@ Notes on Cursor flags:
 [Reviewer feedback]
 ```
 
-3. Check verdict:
+4. Check verdict:
    - **VERDICT: APPROVED** -> proceed to Phase 4 Step 6 (commit & approve)
    - **VERDICT: REVISE** -> go to Step 5
    - No clear verdict but positive / no actionable items -> treat as approved
+   - Helper state `completed-empty-output` -> treat as failed review attempt, surface stderr/status, fix invocation or prompt handling, then retry
+   - Helper state `needs-operator-decision` -> surface status log, note any `stall-warning` heartbeat lines as non-terminal operator hints, and decide whether to keep waiting, abort, or retry with different helper parameters
    - Max rounds (`MAX_ROUNDS`) reached -> present to user for manual decision (proceed or stop)
 
 #### Step 5: Address Feedback & Re-verify
@@ -309,6 +360,8 @@ If a revision contradicts the user's explicit requirements, skip it and note it
 
 #### Step 6: Re-submit to Reviewer (Rounds 2-N)
 
+Rewrite `/tmp/milestone-review-${REVIEW_ID}.sh` for the next round. The script should contain the reviewer invocation only; do not run it directly.
+
 **If `REVIEWER_CLI` is `codex`:**
 
 Resume the existing session:
@@ -367,20 +420,31 @@ Changes made:
 Re-review. If solid, end with: VERDICT: APPROVED
 If more changes needed, end with: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
-
-jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
 If resume fails, fall back to fresh `cursor-agent -p` with context about prior rounds.
 
+Do not run `jq` extraction until after the helper or fallback execution completes, then extract `/tmp/milestone-review-${REVIEW_ID}.md` from the JSON response.
+
+After updating `/tmp/milestone-review-${REVIEW_ID}.sh`, run the same helper/fallback flow from Round 1.
+
 Return to Step 4.
 
 #### Step 7: Cleanup Per Milestone
 
 ```bash
-rm -f /tmp/milestone-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.json
+rm -f \
+  /tmp/milestone-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.json \
+  /tmp/milestone-review-${REVIEW_ID}.stderr \
+  /tmp/milestone-review-${REVIEW_ID}.status \
+  /tmp/milestone-review-${REVIEW_ID}.runner.out \
+  /tmp/milestone-review-${REVIEW_ID}.sh
 ```
 
+If the round failed, produced empty output, or reached operator-decision timeout, keep `.stderr`, `.status`, and `.runner.out` until the issue is diagnosed instead of deleting them immediately.
+
 ### Phase 6: Completion (REQUIRED SUB-SKILL)
 
 After all milestones are approved and committed:
diff --git a/skills/implement-plan/opencode/SKILL.md b/skills/implement-plan/opencode/SKILL.md
index 13d1a43..8efb202 100644
--- a/skills/implement-plan/opencode/SKILL.md
+++ b/skills/implement-plan/opencode/SKILL.md
@@ -154,7 +154,20 @@ Do NOT push. After committing:
 REVIEW_ID=$(uuidgen | tr '[:upper:]' '[:lower:]' | head -c 8)
 ```
 
-Use for all temp file paths: `/tmp/milestone-${REVIEW_ID}.md` and `/tmp/milestone-review-${REVIEW_ID}.md`.
+Use `REVIEW_ID` for all milestone review temp file paths:
+- `/tmp/milestone-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.md`
+- `/tmp/milestone-review-${REVIEW_ID}.json`
+- `/tmp/milestone-review-${REVIEW_ID}.stderr`
+- `/tmp/milestone-review-${REVIEW_ID}.status`
+- `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+- `/tmp/milestone-review-${REVIEW_ID}.sh`
+
+Resolve the shared runtime helper path before writing the command script:
+
+```bash
+REVIEWER_RUNTIME=~/.config/opencode/skills/reviewer-runtime/run-review.sh
+```
 
 #### Step 2: Write Review Payload
 
@@ -183,6 +196,13 @@ Write to `/tmp/milestone-${REVIEW_ID}.md`:
 
 #### Step 3: Submit to Reviewer (Round 1)
 
+Write the reviewer invocation to `/tmp/milestone-review-${REVIEW_ID}.sh` as a bash script:
+
+```bash
+#!/usr/bin/env bash
+set -euo pipefail
+```
+
 **If `REVIEWER_CLI` is `codex`:**
 
 ```bash
@@ -203,7 +223,7 @@ Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE"
 ```
 
-Capture the Codex session ID from output (line `session id: <uuid>`). Store as `CODEX_SESSION_ID`.
+Do not try to capture the Codex session ID yet. When using the helper, extract it from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the command completes (look for `session id: <uuid>`), then store it as `CODEX_SESSION_ID` for resume in subsequent rounds.
 
 **If `REVIEWER_CLI` is `claude`:**
 
@@ -221,8 +241,7 @@ Evaluate:
 Be specific and actionable. If solid, end with exactly: VERDICT: APPROVED
 If changes are needed, end with exactly: VERDICT: REVISE" \
   --model ${REVIEWER_MODEL} \
-  --allowedTools Read \
-  > /tmp/milestone-review-${REVIEW_ID}.md
+  --allowedTools Read
 ```
 
 **If `REVIEWER_CLI` is `cursor`:**
@@ -247,19 +266,43 @@ If changes are needed, end with exactly: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
 ```
 
-Extract session ID and review text (requires `jq`):
+For `cursor`, the command script writes raw JSON to `/tmp/milestone-review-${REVIEW_ID}.json`. Do not run `jq` extraction until after the helper or fallback execution completes. If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivalent.
+
+Run the command script through the shared helper when available:
+
+```bash
+if [ -x "$REVIEWER_RUNTIME" ]; then
+  "$REVIEWER_RUNTIME" \
+    --command-file /tmp/milestone-review-${REVIEW_ID}.sh \
+    --stdout-file /tmp/milestone-review-${REVIEW_ID}.runner.out \
+    --stderr-file /tmp/milestone-review-${REVIEW_ID}.stderr \
+    --status-file /tmp/milestone-review-${REVIEW_ID}.status
+else
+  echo "Warning: reviewer runtime helper not found at $REVIEWER_RUNTIME; falling back to direct synchronous review." >&2
+  bash /tmp/milestone-review-${REVIEW_ID}.sh >/tmp/milestone-review-${REVIEW_ID}.runner.out 2>/tmp/milestone-review-${REVIEW_ID}.stderr
+fi
+```
+
+After the command completes:
+- If `REVIEWER_CLI=cursor`, extract the final review text:
 
 ```bash
 CURSOR_SESSION_ID=$(jq -r '.session_id' /tmp/milestone-review-${REVIEW_ID}.json)
 jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
-If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivalent.
+- If `REVIEWER_CLI=codex`, extract `CODEX_SESSION_ID` from `/tmp/milestone-review-${REVIEW_ID}.runner.out` after the helper or fallback run. If the review text is only in `.runner.out`, move or copy the actual review body into `/tmp/milestone-review-${REVIEW_ID}.md` before verdict parsing.
+
+Fallback is allowed only when the helper is missing or not executable.
 
 #### Step 4: Read Review & Check Verdict
 
 1. Read `/tmp/milestone-review-${REVIEW_ID}.md`
-2. Present review to the user:
+2. If the review failed, produced empty output, or reached helper timeout, also read:
+   - `/tmp/milestone-review-${REVIEW_ID}.stderr`
+   - `/tmp/milestone-review-${REVIEW_ID}.status`
+   - `/tmp/milestone-review-${REVIEW_ID}.runner.out`
+3. Present review to the user:
 
 ```
 ## Milestone Review — Round N (reviewer: ${REVIEWER_CLI} / ${REVIEWER_MODEL})
@@ -267,10 +310,12 @@ If `jq` is not installed, inform the user: `brew install jq` (macOS) or equivale
 [Reviewer feedback]
 ```
 
-3. Check verdict:
+4. Check verdict:
    - **VERDICT: APPROVED** -> proceed to Phase 5 Step 6 (commit & approve)
    - **VERDICT: REVISE** -> go to Step 5
    - No clear verdict but positive / no actionable items -> treat as approved
+   - Helper state `completed-empty-output` -> treat as failed review attempt, surface stderr/status, fix invocation or prompt handling, then retry
+   - Helper state `needs-operator-decision` -> surface status log, note any `stall-warning` heartbeat lines as non-terminal operator hints, and decide whether to keep waiting, abort, or retry with different helper parameters
    - Max rounds (`MAX_ROUNDS`) reached -> present to user for manual decision (proceed or stop)
 
 #### Step 5: Address Feedback & Re-verify
@@ -290,6 +335,8 @@ If a revision contradicts the user's explicit requirements, skip it and note it
 
 #### Step 6: Re-submit to Reviewer (Rounds 2-N)
 
+Rewrite `/tmp/milestone-review-${REVIEW_ID}.sh` for the next round. The script should contain the reviewer invocation only; do not run it directly.
+
 **If `REVIEWER_CLI` is `codex`:**
 
 Resume the existing session:
@@ -348,20 +395,31 @@ Changes made:
 Re-review. If solid, end with: VERDICT: APPROVED
 If more changes needed, end with: VERDICT: REVISE" \
   > /tmp/milestone-review-${REVIEW_ID}.json
-
-jq -r '.result' /tmp/milestone-review-${REVIEW_ID}.json > /tmp/milestone-review-${REVIEW_ID}.md
 ```
 
 If resume fails, fall back to fresh `cursor-agent -p` with context about prior rounds.
 
+Do not run `jq` extraction until after the helper or fallback execution completes, then extract `/tmp/milestone-review-${REVIEW_ID}.md` from the JSON response.
+
+After updating `/tmp/milestone-review-${REVIEW_ID}.sh`, run the same helper/fallback flow from Round 1.
+
 Return to Step 4.
 
 #### Step 7: Cleanup Per Milestone
 
 ```bash
-rm -f /tmp/milestone-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.md /tmp/milestone-review-${REVIEW_ID}.json
+rm -f \
+  /tmp/milestone-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.md \
+  /tmp/milestone-review-${REVIEW_ID}.json \
+  /tmp/milestone-review-${REVIEW_ID}.stderr \
+  /tmp/milestone-review-${REVIEW_ID}.status \
+  /tmp/milestone-review-${REVIEW_ID}.runner.out \
+  /tmp/milestone-review-${REVIEW_ID}.sh
 ```
 
+If the round failed, produced empty output, or reached operator-decision timeout, keep `.stderr`, `.status`, and `.runner.out` until the issue is diagnosed instead of deleting them immediately.
+
 ### Phase 7: Completion (REQUIRED SUB-SKILL)
 
 After all milestones are approved and committed: