Implementation Review Mode

Read [workflow.md](workflow.md) for detailed phases and anti-patterns.

Conduct a John Carmack-level review of implementation changes on the current branch.

Role: Code Review Coordinator (NOT the reviewer) Backends: OpenCode (opencode) or RepoPrompt (rp)

⚠️ RepoPrompt 1.6.0+ Required (rp backend): The RP backend uses builder review mode. Check version: rp-cli --version.

CRITICAL: flowctl is BUNDLED — NOT installed globally. which flowctl will fail (expected). Always use:

ROOT="$(git rev-parse --show-toplevel)"
OPENCODE_DIR="$ROOT/.opencode"
FLOWCTL="$OPENCODE_DIR/bin/flowctl"

Backend Selection

Priority (first match wins):

1. --review=rp|opencode|export|none argument
2. FLOWREVIEWBACKEND env var (rp, opencode, none)
3. .flow/config.json → review.backend
4. Interactive prompt if rp-cli available (and not in Ralph mode)
5. Default: opencode

Parse from arguments first

Check $ARGUMENTS for:

• --review=rp or --review rp → use rp
• --review=opencode or --review opencode → use opencode
• --review=export or --review export → use export
• --review=none or --review none → skip review

If found, use that backend and skip all other detection.

Otherwise detect

# Check available backends
HAVE_RP=0;
if command -v rp-cli >/dev/null 2>&1; then
  HAVE_RP=1;
elif [[ -x /opt/homebrew/bin/rp-cli || -x /usr/local/bin/rp-cli ]]; then
  HAVE_RP=1;
fi;

# Get configured backend
BACKEND="${FLOW_REVIEW_BACKEND:-}";
if [[ -z "$BACKEND" ]]; then
  BACKEND="$($FLOWCTL config get review.backend --json 2>/dev/null | jq -r '.value // empty')";
fi

MUST RUN the detection command above and use its result. Do not assume rp-cli is missing without running it.

If no backend configured and rp available

If BACKEND is empty AND HAVERP=1, AND not in Ralph mode (FLOWRALPH not set):

Output this question as text:

Which review backend?
a) OpenCode review (GPT-5.2, reasoning high)
b) RepoPrompt (macOS, visual builder)

(Reply: "a", "opencode", "b", "rp", or just tell me)

IMPORTANT: Ask this in plain text only. Do NOT use the question tool.

Wait for response. Parse naturally.

Default if empty/ambiguous: opencode

If only one available or in Ralph mode

# Fallback to available
if [[ -z "$BACKEND" ]]; then
  if [[ "$HAVE_RP" == "1" ]]; then BACKEND="opencode"
  else BACKEND="opencode"; fi
fi

Critical Rules

For rp backend:

1. DO NOT REVIEW CODE YOURSELF - you coordinate, RepoPrompt reviews
2. MUST WAIT for actual RP response - never simulate/skip the review
3. MUST use setup-review - handles window selection + builder atomically
4. DO NOT add --json flag to chat-send - it suppresses the review response
5. Re-reviews MUST stay in SAME chat - omit --new-chat after first review

For opencode backend:

1. Use $FLOWCTL opencode impl-review command
2. Pass --base for diff scope and --receipt for session continuity
3. Parse verdict from command output

For all backends:

• If REVIEWRECEIPTPATH set: write receipt after review (any verdict)
• Any failure → output <promise>RETRY</promise> and stop

FORBIDDEN:

• Self-declaring SHIP without actual backend verdict
• Mixing backends mid-review (stick to one)
• Skipping review when backend is "none" without user consent

Input

Arguments: $ARGUMENTS Format: [focus areas or task ID]

Reviews all changes on current branch vs main/master.

Workflow

See [workflow.md](workflow.md) for full details on each backend.

ROOT="$(git rev-parse --show-toplevel)"
OPENCODE_DIR="$ROOT/.opencode"
FLOWCTL="$OPENCODE_DIR/bin/flowctl"
REPO_ROOT="$(git rev-parse --show-toplevel 2>/dev/null || pwd)"

Step 0: Detect Backend

Run backend detection from SKILL.md above. Then branch:

OpenCode Backend

TASK_ID="${1:-}"
BASE_COMMIT="${BASE_COMMIT:-main}"
RECEIPT_PATH="${REVIEW_RECEIPT_PATH:-/tmp/impl-review-receipt.json}"

$FLOWCTL opencode impl-review "$TASK_ID" --base "$BASE_COMMIT" --receipt "$RECEIPT_PATH"
# Output includes VERDICT=SHIP|NEEDS_WORK|MAJOR_RETHINK

On NEEDS_WORK: fix code, commit, re-run (receipt enables session continuity).

RepoPrompt Backend

# Step 1: Identify changes
git branch --show-current
git log main..HEAD --oneline 2>/dev/null || git log master..HEAD --oneline
git diff main..HEAD --name-only 2>/dev/null || git diff master..HEAD --name-only

# Step 2: Atomic setup
eval "$($FLOWCTL rp setup-review --repo-root "$REPO_ROOT" --summary "Review implementation: <summary>")"
# Outputs W=<window> T=<tab>. If fails → <promise>RETRY</promise>

# Step 3: Augment selection
$FLOWCTL rp select-add --window "$W" --tab "$T" path/to/changed/files...

# Step 4: Build and send review prompt (see workflow.md)
$FLOWCTL rp chat-send --window "$W" --tab "$T" --message-file /tmp/review-prompt.md --new-chat --chat-name "Impl Review: [BRANCH]"

# Step 5: Write receipt if REVIEW_RECEIPT_PATH set

Fix Loop (INTERNAL - do not exit to Ralph)

If verdict is NEEDS_WORK, loop internally until SHIP:

1. Parse issues from reviewer feedback (Critical → Major → Minor)
2. Fix code and run tests/lints
3. Commit fixes (mandatory before re-review)
4. Re-review:

- OpenCode: $FLOWCTL opencode impl-review "$TASKID" --base "$BASECOMMIT" --receipt "$RECEIPT_PATH" - RP: $FLOWCTL rp chat-send --window "$W" --tab "$T" --message-file /tmp/re-review.md (NO --new-chat)

1. Repeat until <verdict>SHIP</verdict>

CRITICAL: For RP, re-reviews must stay in the SAME chat so reviewer has context. Only use --new-chat on the FIRST review.