@hanzlaa/rcode 3.6.16 → 4.1.0

package/rcode/workflows/plan.md

@@ -0,0 +1,1062 @@
+<purpose>
+Create executable phase prompts (SPRINT.md files) for a roadmap phase with integrated research and verification. Default flow: Research (if needed) -> Plan -> Verify -> Done. Orchestrates rcode-phase-researcher, rcode-planner, and rcode-sprint-checker agents with a revision loop (max 3 iterations).
+</purpose>
+
+<output_format>
+Open with banner:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► PLANNING PHASE {NN}
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
+
+TaskCreate at start:
+- TaskCreate: "Load phase scope and context"
+- TaskCreate: "Research phase (if enabled)"
+- TaskCreate: "Spawn rcode-planner → SPRINT.md"
+- TaskCreate: "Run rcode-sprint-checker verification"
+- TaskCreate: "Revise plan (up to 3 iterations)" — only if checker flags issues
+- TaskCreate: "Commit SPRINT.md + update state"
+
+Spawning indicators:
+```
+◆ Spawning rcode-phase-researcher...
+✓ Research complete: RESEARCH.md ({N} lines)
+
+◆ Spawning rcode-planner...
+✓ Planner complete: SPRINT.md ({N} stories, {M} points)
+
+◆ Spawning rcode-sprint-checker...
+✓ Check complete: {PASS|PARTIAL|FAIL} — see CHECK.md
+```
+
+Closure:
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► PLAN READY ✓  ({N} stories, {M} points)
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
+End with Next Up routing to /rcode-execute.
+</output_format>
+
+<required_reading>
+@.rcode/references/auto-init-guard.md
+@.rcode/references/output-format.md
+Read all files referenced by the invoking prompt's execution_context before starting.
+
+<!-- ui-brand.md (254 lines): only load when phase goal/CONTEXT.md contains UI signals (frontend|ui|component|design|style|brand) -->
+${PHASE_GOAL_HAS_UI ? '@.rcode/references/ui-brand.md' : ''}
+@.rcode/references/revision-loop.md
+@.rcode/references/gate-prompts.md
+@.rcode/references/agent-contracts.md
+@.rcode/references/gates.md
+@.rcode/references/karpathy-guidelines.md
+@.rcode/references/thinking-models-planning.md
+</required_reading>
+
+<available_agent_types>
+Valid rcode subagent types (use exact names — do not fall back to 'general-purpose'):
+- rcode-phase-researcher — Researches technical approaches for a phase
+- rcode-planner — Creates detailed plans from phase scope
+- rcode-sprint-checker — Reviews plan quality before execution
+</available_agent_types>
+
+<process>
+
+## 1. Initialize
+
+Load all context in one call (paths only to minimize orchestrator context):
+
+```bash
+INIT=$(node ".rcode/bin/rcode-tools.cjs" init sprint-plan "$PHASE" 2>/dev/null)
+if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_RESEARCHER=$(node ".rcode/bin/rcode-tools.cjs" agent-skills rcode-phase-researcher 2>/dev/null || echo "")
+AGENT_SKILLS_PLANNER=$(node ".rcode/bin/rcode-tools.cjs" agent-skills rcode-planner 2>/dev/null || echo "")
+AGENT_SKILLS_CHECKER=$(node ".rcode/bin/rcode-tools.cjs" agent-skills rcode-sprint-checker 2>/dev/null || echo "")
+CONTEXT_WINDOW=$(node ".rcode/bin/rcode-tools.cjs" config-get context_window 2>/dev/null)
+CONTEXT_WINDOW=${CONTEXT_WINDOW:-200000}  # config-get exits 0 with empty output when key absent; || fallback won't fire
+
+# Detect UI signals in phase goal + CONTEXT.md to decide whether to load ui-brand.md (254 lines)
+PHASE_GOAL_HAS_UI=$(grep -iEl "frontend|ui|component|design|style|brand" \
+  .planning/phases/*${PHASE_NUMBER}*/*-CONTEXT.md \
+  .planning/ROADMAP.md 2>/dev/null | head -1)
+```
+
+If `INIT` is empty, or `INIT.ok` is false or absent (null/undefined — `init sprint-plan` may omit the key), print error and exit:
+```
+Error: rcode-tools init failed. Verify .rcode/ is installed and state.json is valid.
+```
+
+When `CONTEXT_WINDOW >= 500000`, the planner prompt includes prior phase CONTEXT.md files so cross-phase decisions are consistent (e.g., "use library X for all data fetching" from Phase 2 is visible to Phase 5's planner).
+
+Parse JSON for: `researcher_model`, `planner_model`, `checker_model`, `research_enabled`, `plan_checker_enabled`, `nyquist_validation_enabled`, `commit_docs`, `text_mode`, `phase_found`, `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `padded_phase`, `has_research`, `has_context`, `has_reviews`, `has_plans`, `plan_count`, `phase_status`, `planning_exists`, `roadmap_exists`, `phase_req_ids`, `response_language`.
+
+**If `response_language` is set:** Include `response_language: {value}` in all spawned subagent prompts so any user-facing output stays in the configured language.
+
+**File paths (for <files_to_read> blocks):** `state_path`, `roadmap_path`, `requirements_path`, `context_path`, `research_path`, `verification_path`, `uat_path`, `reviews_path`. These are null if files don't exist.
+
+**If `planning_exists` is false:** Error — run `/rcode-new-project` first.
+
+## 2. Parse and Normalize Arguments
+
+Extract from $ARGUMENTS: phase number (integer or decimal like `2.1`), flags (`--research`, `--skip-research`, `--gaps`, `--skip-verify`, `--from-stub`, `--prd <filepath>`, `--reviews`, `--text`).
+
+Set `TEXT_MODE=true` if `--text` is present in $ARGUMENTS OR `text_mode` from init JSON is `true`. When `TEXT_MODE` is active, replace every `AskUserQuestion` call with a plain-text numbered list and ask the user to type their choice number. This is required for Claude Code remote sessions (`/rc` mode) where TUI menus don't work through the Claude App.
+
+Extract `--prd <filepath>` from $ARGUMENTS. If present, set PRD_FILE to the filepath.
+
+**Detect gaps mode:**
+```bash
+if [[ "$ARGUMENTS" =~ (^|[[:space:]])--gaps($|[[:space:]]) ]]; then
+  GAPS_MODE=true
+else
+  GAPS_MODE=false
+fi
+```
+
+When `GAPS_MODE=true`, the workflow switches to **gap-closure planning**: read the phase's VERIFICATION.md, extract verification gaps classified `gap_found` or `partial`, and produce a single new numbered plan file (`NNN-NN-SPRINT.md`) that closes them. Research, CONTEXT.md gating, and VALIDATION.md creation are skipped — gaps are grounded in already-shipped code, not new design work.
+
+**Detect from-stub mode (closes #736):**
+```bash
+if [[ "$ARGUMENTS" =~ (^|[[:space:]])--from-stub($|[[:space:]]) ]]; then
+  FROM_STUB_MODE=true
+else
+  FROM_STUB_MODE=false
+fi
+```
+
+When `FROM_STUB_MODE=true`, the workflow reads an existing stub `SPRINT.md` (or any `*-SPRINT.md`) already present in the phase directory and treats it as the authoritative task list — the researcher and CONTEXT.md gating are skipped. The stub is passed to the planner as `existing_stub_content` so it can refine, expand, and add implementation detail without rewriting the structure. This is the correct flow when a user has partially sketched a plan by hand or a prior workflow run created a skeleton.
+
+**From-stub resolution:**
+```bash
+if [[ "$FROM_STUB_MODE" == "true" ]]; then
+  STUB_FILE=$(ls "${PHASE_DIR}"/*-SPRINT.md 2>/dev/null | head -1)
+  if [[ -z "$STUB_FILE" ]]; then
+    echo "Error: --from-stub requires an existing SPRINT.md in the phase directory."
+    echo "Found: ${PHASE_DIR}"
+    echo "  (create a stub manually then re-run with --from-stub)"
+    exit 1
+  fi
+  STUB_CONTENT=$(cat "$STUB_FILE")
+  echo "◆ From-stub mode: using $(basename $STUB_FILE) as planner input"
+fi
+```
+
+When `FROM_STUB_MODE=true`: skip steps 4 (CONTEXT.md), 5 (Research), 5.5 (Validation strategy). Jump directly to step 8 (Spawn rcode-planner). Pass `STUB_CONTENT` and `STUB_FILE` to the planner prompt so it refines rather than replaces the stub.
+
+**If no phase number:** Detect next unplanned phase from roadmap.
+
+**If `phase_found` is false:** Validate phase exists in ROADMAP.md. If valid, create the directory using `phase_slug` and `padded_phase` from init:
+```bash
+mkdir -p ".planning/phases/${padded_phase}-${phase_slug}"
+```
+
+**Existing artifacts from init:** `has_research`, `has_plans`, `plan_count`.
+
+**TASKS.md ingestion (#385 chain).** If the phase directory contains a `TASKS.md` file (typically auto-extracted by `/rcode-add-phase` from a bulk `/rcode-quick` or `/rcode-do` route), read it now:
+
+```bash
+TASKS_FILE=".planning/phases/${padded_phase}-${phase_slug}/TASKS.md"
+HAS_TASKS=$([ -f "$TASKS_FILE" ] && echo true || echo false)
+```
+
+When `HAS_TASKS=true`:
+- Pass the TASKS.md content to the planner agent as authoritative phase scope. The planner uses it as the input list — each entry becomes a candidate sprint task in SPRINT.md.
+- Surface this in the opening banner: *"Phase scope source: TASKS.md ({N} entries auto-extracted from bulk route on {date})"*.
+- Do NOT re-prompt the user for scope when TASKS.md is present — they already provided the list once at the /rcode-quick or /rcode-do entry point. The whole point of the auto-route chain is that the user doesn't paste the same content multiple times.
+
+## 2.5. Validate `--reviews` Prerequisite
+
+**Skip if:** No `--reviews` flag.
+
+**If `--reviews` AND `--gaps`:** Error — cannot combine `--reviews` with `--gaps`. These are conflicting modes.
+
+**If `--reviews` AND `has_reviews` is false (no REVIEWS.md in phase dir):**
+
+Error:
+```
+No REVIEWS.md found for Phase {N}. Run reviews first:
+
+/rcode-review --phase {N}
+
+Then re-run /rcode-plan {N} --reviews
+```
+Exit workflow.
+
+## 3. Validate Phase
+
+```bash
+PHASE_INFO=$(node ".rcode/bin/rcode-tools.cjs" roadmap get-phase "${PHASE}")
+```
+
+**If `found` is false:** Error with available phases. **If `found` is true:** Extract `phase_number`, `phase_name`, `goal` from JSON.
+
+
+@.rcode/workflows/plan-prd-express.md
+
+
+## 3.6. Handle `--gaps` Mode
+
+**Skip unless:** `GAPS_MODE=true`.
+
+**Purpose:** Read `NNN-VERIFICATION.md`, extract failing/partial gaps, count existing plan files, and prepare a `gap_list` payload to feed the planner. On completion, control flow continues at step 8 (skipping CONTEXT.md gating, research, and validation-strategy creation).
+
+**Step 1: Locate VERIFICATION.md**
+
+```bash
+PHASE_DIR=$(node ".rcode/bin/rcode-tools.cjs" roadmap get-phase "${PHASE}" --pick dir 2>/dev/null || echo "")
+# Fallback if --pick dir not supported. TODO(#118): expose roadmap --pick dir cleanly.
+if [[ -z "$PHASE_DIR" ]]; then
+  PHASE_DIR=$(ls -d .planning/phases/${padded_phase}-* 2>/dev/null | head -1)
+fi
+
+VERIFICATION_FILE=$(ls "${PHASE_DIR}"/*-VERIFICATION.md 2>/dev/null | head -1)
+```
+
+**If `VERIFICATION_FILE` is empty:**
+```
+Error: No VERIFICATION.md found for Phase {X}. Gap-closure planning requires the phase to have run through the verifier first.
+
+Try:
+  /rcode-execute {X} ${RCODE_WS}      # run or re-run execution + verification
+```
+Exit workflow.
+
+**Step 2: Extract gaps from VERIFICATION.md**
+
+Parse the file for gap entries with `status: gap_found` or `status: partial`. Inspect these sections:
+- `## Automated Gap` (or `## Automated Gaps`)
+- `## Human Verification Required`
+- Any findings block that includes a `status:` field set to `gap_found` or `partial`
+
+Collect into `GAP_LIST` (an ordered list where each entry has: id, title, expected, actual, status, source_section, severity if present).
+
+If `GAP_LIST` is empty, display:
+```
+Phase {X} VERIFICATION.md contains no gap_found or partial items — nothing to close.
+Report: {VERIFICATION_FILE}
+```
+Exit workflow.
+
+**Step 3: Determine next plan number**
+
+```bash
+EXISTING_PLAN_COUNT=$(ls "${PHASE_DIR}"/*-SPRINT.md 2>/dev/null | wc -l | tr -d ' ')
+# Issue #652 — no leading zeros in planning artifacts. Phase 8 not 08, plan 2 not 02.
+NEXT_PLAN_NUMBER=$((EXISTING_PLAN_COUNT + 1))
+PADDED_PHASE="${PHASE}"
+GAP_PLAN_FILENAME="${PADDED_PHASE}-${NEXT_PLAN_NUMBER}-SPRINT.md"
+GAP_PLAN_PATH="${PHASE_DIR}/${GAP_PLAN_FILENAME}"
+```
+
+If `EXISTING_PLAN_COUNT == 0`, there is no prior execution to reference. Display a warning but proceed — the planner can still close verification gaps.
+
+**Step 4: Gather prior plans for planner context**
+
+```bash
+EXISTING_PLAN_FILES=$(ls "${PHASE_DIR}"/*-SPRINT.md 2>/dev/null | tr '\n' ' ')
+EXISTING_SUMMARY_FILES=$(ls "${PHASE_DIR}"/*-SUMMARY.md 2>/dev/null | tr '\n' ' ')
+```
+
+**Step 5: Display banner**
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► GAP-CLOSURE PLANNING — Phase {X}
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Verification report: {VERIFICATION_FILE}
+Gaps to close:       {count(GAP_LIST)}
+Existing plans:      {EXISTING_PLAN_COUNT}
+New plan file:       {GAP_PLAN_FILENAME}
+```
+
+**Step 6: Skip ahead**
+
+Control flow jumps directly to step 8 (Spawn rcode-planner). Steps 4 (CONTEXT.md), 5 (Research), and 5.5 (Validation) are ALL skipped when `GAPS_MODE=true`.
+
+Step 8 will consume these variables when filling the planner prompt:
+- `GAP_LIST` — serialized list of gaps (id, title, expected, actual, status)
+- `GAP_PLAN_PATH` — exact output path the planner must write
+- `EXISTING_PLAN_FILES` / `EXISTING_SUMMARY_FILES` — prior phase context
+- `VERIFICATION_FILE` — authoritative source-of-truth
+
+After the planner returns, the existing plan-checker / revision loop (step 10 onward) runs unchanged — gap plans are verified just like normal plans.
+
+## 4. Load CONTEXT.md
+
+**Skip if:** PRD express path was used (CONTEXT.md already created in step 3.5) OR `GAPS_MODE=true` (gap closure is grounded in VERIFICATION.md, not CONTEXT.md).
+
+Check `context_path` from init JSON.
+
+If `context_path` is not null, display: `Using phase context from: ${context_path}`
+
+**If `context_path` is null (no CONTEXT.md exists):**
+
+Read discuss mode for context gate label:
+```bash
+DISCUSS_MODE=$(node ".rcode/bin/rcode-tools.cjs" config-get workflow.discuss_mode 2>/dev/null)
+DISCUSS_MODE=${DISCUSS_MODE:-discuss}  # config-get exits 0 with empty output when key absent
+```
+
+If `TEXT_MODE` is true, present as a plain-text numbered list:
+```
+No CONTEXT.md found for Phase {X}. Plans will use research and requirements only — your design preferences won't be included.
+
+1. Continue without context — Plan using research + requirements only
+[If DISCUSS_MODE is "assumptions":]
+2. Gather context (assumptions mode) — Analyze codebase and surface assumptions before planning
+[If DISCUSS_MODE is "discuss" or unset:]
+2. Run discuss-phase first — Capture design decisions before planning
+
+Enter number:
+```
+
+Otherwise use AskUserQuestion:
+- header: "No context"
+- question: "No CONTEXT.md found for Phase {X}. Plans will use research and requirements only — your design preferences won't be included. Continue or capture context first?"
+- options:
+  - "Continue without context" — Plan using research + requirements only
+  If `DISCUSS_MODE` is `"assumptions"`:
+  - "Gather context (assumptions mode)" — Analyze codebase and surface assumptions before planning
+  If `DISCUSS_MODE` is `"discuss"` (or unset):
+  - "Run discuss-phase first" — Capture design decisions before planning
+
+If "Continue without context": Proceed to step 5.
+If "Run discuss-phase first":
+  **IMPORTANT:** Do NOT invoke discuss-phase as a nested Skill/Task call — AskUserQuestion
+  does not work correctly in nested subcontexts (#1009). Instead, display the command
+  and exit so the user runs it as a top-level command:
+  ```
+  Run this command first, then re-run /rcode-plan {X} ${RCODE_WS}:
+
+  /rcode-discuss-phase {X} ${RCODE_WS}
+  ```
+  **Exit the sprint-plan workflow. Do not continue.**
+
+
+@.rcode/workflows/plan-research-validation.md
+
+
+## 6. Check Existing Plans
+
+```bash
+ls "${PHASE_DIR}"/*-SPRINT.md 2>/dev/null || true
+```
+
+**If exists AND `--reviews` flag:** Skip prompt — go straight to replanning (the purpose of `--reviews` is to replan with review feedback).
+
+**If exists AND no `--reviews` flag:** Ask the user what they'd like to do. Tailor the message to context — do NOT say "as per the workflow" or expose implementation details. Examples:
+
+- If `phase_status` is `complete` or `executed`:
+  > "Phase {N} ({name}) already shipped {plan_count} plans and is marked {status}. Do you want to review those plans, add more, or replan from scratch?"
+
+- If `phase_status` is `in_progress` or `planned` (or null):
+  > "Phase {N} ({name}) already has {plan_count} plan(s). Want to add more, review what's there, or start fresh?"
+
+Always offer exactly three numbered options:
+1. Add more plans
+2. View existing plans
+3. Replan from scratch
+
+Wait for the user's choice before proceeding. Do not auto-select.
+
+**If user picks option 1 (Add more plans) — issue #650:**
+
+This is **NOT** a license to hand-write a new SPRINT.md inline. Continue down the
+normal pipeline exactly as if no plans existed yet:
+
+1. Proceed to Step 7 (context-paths) and Step 7.5 (Nyquist verification) as normal.
+2. Spawn `rcode-planner` via `@.rcode/workflows/plan-spawn-planner.md` (Step 8). The
+   planner subagent is mandatory — the orchestrator must NOT write SPRINT.md
+   directly via the `Write` tool. Pass the existing plan list to the planner so
+   it picks the next plan number and avoids re-covering shipped tasks.
+3. After the planner returns, run sprint-checker (Step 10) the same as a
+   first-time plan. The "PLANNED ✓" banner is gated on a passing CHECK.md.
+
+A run that emits a SPRINT.md without a corresponding planner Task() invocation
+in the same turn is a malfunction — see issue #650. Stop and report instead of
+shipping a hand-rolled plan.
+
+**If user picks option 3 (Replan from scratch):**
+
+Same as option 1, but pass the existing plans to the planner with a `replace:
+true` directive. Existing PLAN.md files are renamed to `*-SUPERSEDED.md` (do
+not delete) before the planner writes the new ones. Subagent invocation is
+still mandatory.
+
+**If user picks option 2 (View existing plans):**
+
+Display a sprint summary table (sprint id → one-line goal).
+
+Then run a **best-effort codebase overlap check** before showing the execute prompt — Closes #596.
+
+**This check is always informational. It never blocks, never errors, never fails the workflow.** If any step below cannot complete for any reason, skip it silently and proceed straight to the execute prompt.
+
+1. Read the SPRINT.md files for this phase (they are already on disk — no tool calls needed beyond `Read`).
+2. From each sprint's `files_modified:` frontmatter list, note which paths already exist on disk vs. which are new.
+3. Separately, look at the sprint *goals* and compare against modules/components the codebase already has. Use your knowledge from any files already read this session; do NOT spawn new reads just for this check.
+4. Report what you found — one compact block:
+
+```
+Codebase overlap check (best-effort):
+  ✓ N files already exist — plans will extend them
+  + M files are new — will be created
+  ⚠ Possible overlap: [file A] in the codebase may already cover [sprint X goal] — worth checking before executing
+```
+
+If nothing notable: one line — `No obvious conflicts detected.`
+
+**Hard rules (dead-ends — nothing here can cause failure):**
+- If a SPRINT.md can't be read → skip it, don't error
+- If files_modified is empty or absent → skip the file check, move on
+- If you're uncertain whether an overlap is real → don't mention it (false positives are noise)
+- If the whole check produces nothing → omit the block entirely, go straight to execute prompt
+- **Never ask a follow-up question about the overlap** — state it and move on
+- **Never refuse to show the execute prompt** because of an overlap finding
+
+Only after showing overlap results (or skipping them), show the execute prompt.
+
+## 7. Use Context Paths from INIT
+
+Extract from INIT JSON:
+
+```bash
+_rcode_field() { node -e "const o=JSON.parse(process.argv[1]); const v=o[process.argv[2]]; process.stdout.write(v==null?'':String(v))" "$1" "$2"; }
+STATE_PATH=$(_rcode_field "$INIT" state_path)
+ROADMAP_PATH=$(_rcode_field "$INIT" roadmap_path)
+REQUIREMENTS_PATH=$(_rcode_field "$INIT" requirements_path)
+RESEARCH_PATH=$(_rcode_field "$INIT" research_path)
+VERIFICATION_PATH=$(_rcode_field "$INIT" verification_path)
+UAT_PATH=$(_rcode_field "$INIT" uat_path)
+CONTEXT_PATH=$(_rcode_field "$INIT" context_path)
+REVIEWS_PATH=$(_rcode_field "$INIT" reviews_path)
+```
+
+## 7.5. Verify Nyquist Artifacts
+
+Skip if `nyquist_validation_enabled` is false OR `research_enabled` is false.
+
+Also skip if all of the following are true:
+- `research_enabled` is false
+- `has_research` is false
+- no `--research` flag was provided
+
+In that no-research path, Nyquist artifacts are **not required** for this run.
+
+```bash
+VALIDATION_EXISTS=$(ls "${PHASE_DIR}"/*-VALIDATION.md 2>/dev/null | head -1)
+```
+
+If missing and Nyquist is still enabled/applicable — ask user:
+1. Re-run: `/rcode-plan {PHASE} --research ${RCODE_WS}`
+2. Disable Nyquist with the exact command:
+   `node ".rcode/bin/rcode-tools.cjs" config-set workflow.nyquist_validation false`
+3. Continue anyway (plans fail Dimension 8)
+
+Proceed to Step 8 only if user selects 2 or 3.
+
+
+@.rcode/workflows/plan-spawn-planner.md
+
+## 8.5. File-Ownership & Conflict-Avoidance
+
+After the planner returns SPRINT.md files, run these checks before advancing to step 9.
+These rules were added after overnight parallel builds exposed silent merge-time data loss
+(calorie-calculator-ai, 2026-05-26). The wave-overlap CLI check in step 12.5 catches
+same-wave/same-file issues mechanically; this step catches plan-level ownership gaps earlier.
+
+**Step 1 — Build the cross-sprint file manifest.**
+
+Read each SPRINT.md produced this run and collect its `files_modified:` frontmatter list.
+Produce a de-duplicated map: `file → [sprint_ids that touch it]`.
+
+```bash
+# Quick grep of frontmatter to build the manifest
+grep -A 50 "^files_modified:" "${PHASE_DIR}"/*-SPRINT.md 2>/dev/null
+```
+
+**Step 2 — Flag collisions.**
+
+For each file that appears in 2+ sprint lists:
+
+| Type | Rule |
+|------|------|
+| Two sprints **create** the same file | Plan defect — merge tasks or sequence; only one sprint may create a file |
+| Two same-wave sprints **modify** the same file | Later sprint MUST have `sequential: true` + `sequential_after:` in frontmatter |
+| An aggregator file touched by multiple sprints | Each sprint's `<action>` must use append-only `Edit`, not `Write` |
+
+Aggregator patterns (known high-collision targets):
+- `**/index.ts`, `**/index.tsx` (barrel exports)
+- `**/store/index.ts`, `**/store/index.js`
+- `**/types/index.ts`, `**/types.ts`
+- `main.py`, `app.py`, `router/__init__.py`, `**/__init__.py`
+- `package.json` (scripts / deps blocks)
+
+**Step 3 — Verify-command accuracy.**
+
+Scan every `<verify><automated>` block for raw tool invocations (`tsc --noEmit`, `eslint .`,
+`jest`, `vitest`). If found, check `package.json` scripts and replace with the `pnpm run <script>`
+equivalent. Mismatch example from overnight build: plan wrote `tsc --noEmit` but the project's
+script was named `type-check`.
+
+**Step 4 — Report and gate.**
+
+If any creation collision found → **BLOCK plan acceptance**. Edit the colliding sprint to remove
+the duplicate creation task (have the second sprint import/extend from the first sprint's output).
+
+If any modify-collision found in the same wave → edit the later sprint's frontmatter immediately:
+```yaml
+sequential: true
+sequential_after: <earlier_sprint_id>
+conflicting_files: [<shared_files...>]
+```
+
+Display a summary:
+```
+File-Ownership Check:
+  ✓ N files with single owner
+  ⚠ M aggregator files — append-only instructions verified
+  ✗ K creation collisions (blocked — fixed inline)
+  ~ J sequential flags added
+```
+
+If no issues: `File-Ownership Check: ✓ no collisions.`
+
+**This check is informational for warnings and blocking for creation collisions.**
+It never silently passes a plan where two sprints create the same file.
+
+## 9. Handle Planner Return
+
+- **`## PLANNING COMPLETE`:** Display plan count. If `--skip-verify` or `plan_checker_enabled` is false (from init): skip to step 13. Otherwise: step 10.
+- **`## PHASE SPLIT RECOMMENDED`:** The planner determined the phase is too complex to implement all user decisions without simplifying them. Handle in step 9b.
+- **`## CHECKPOINT REACHED`:** Present to user, get response, spawn continuation (step 12)
+- **`## PLANNING INCONCLUSIVE`:** Show attempts, offer: Add context / Retry / Manual
+
+**Sprint count guard (token cost protection — closes #584):**
+
+After planner returns `## PLANNING COMPLETE`, immediately count sprint files:
+
+```bash
+MAX_SPRINTS=$($TOOL config-get workflow.max_sprints_per_phase 2>/dev/null)
+MAX_SPRINTS=${MAX_SPRINTS:-4}  # config-get exits 0 with empty output when key absent
+SPRINT_COUNT=$(find "${PHASE_DIR}" -maxdepth 1 -name "*-SPRINT.md" | wc -l | tr -d ' ')
+```
+
+If `SPRINT_COUNT > MAX_SPRINTS`:
+
+```
+⚠ Phase {N}: Planner created {SPRINT_COUNT} sprint files (limit: {MAX_SPRINTS}).
+  This phase is too large — the sprint-checker will be expensive and revision
+  loops will multiply the cost.
+
+  Recommended: split this phase into two using /rcode-plan --split {N}
+
+Options:
+  1. Split phase now (recommended)
+  2. Continue anyway (accept higher token cost)
+  3. Re-plan with explicit 4-sprint limit
+```
+
+In `mode: yolo` / autonomous: auto-select option 3 (re-plan with limit). Do not halt or ask.
+
+Re-plan prompt appended: `"IMPORTANT: Create at most {MAX_SPRINTS} SPRINT.md files. Merge smaller tasks into the nearest related sprint instead of creating new ones."`
+
+## 9b. Handle Phase Split Recommendation
+
+When the planner returns `## PHASE SPLIT RECOMMENDED`, it means the phase has too many decisions to implement at full fidelity within the plan budget. The planner proposes groupings.
+
+**Extract from planner return:**
+- Proposed sub-phases (e.g., "17a: processing core (D-01 to D-19)", "17b: billing + config UX (D-20 to D-27)")
+- Which D-XX decisions go in each sub-phase
+- Why the split is necessary (decision count, complexity estimate)
+
+**Present to user:**
+```
+## Phase {X} is too complex for full-fidelity implementation
+
+The planner found {N} decisions that cannot all be implemented without
+simplifying some. Instead of reducing your decisions, we recommend splitting:
+
+**Option 1: Split into sub-phases**
+- Phase {X}a: {name} — {D-XX to D-YY} ({N} decisions)
+- Phase {X}b: {name} — {D-XX to D-YY} ({M} decisions)
+
+**Option 2: Proceed anyway** (planner will attempt all, quality may degrade)
+
+**Option 3: Prioritize** — you choose which decisions to implement now,
+rest become a follow-up phase
+```
+
+Use AskUserQuestion with these 3 options.
+
+**If "Split":** Use `/rcode-insert-phase` to create the sub-phases, then replan each.
+**If "Proceed":** Return to planner with instruction to attempt all decisions at full fidelity, accepting more plans/tasks.
+**If "Prioritize":** Use AskUserQuestion (multiSelect) to let user pick which D-XX are "now" vs "later". Create CONTEXT.md for each sub-phase with the selected decisions.
+
+## 10. Spawn rcode-sprint-checker Agent
+
+Display banner:
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► VERIFYING PLANS
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+◆ Spawning plan checker...
+```
+
+Checker prompt:
+
+```markdown
+<verification_context>
+**Phase:** {phase_number}
+**Phase Goal:** {goal from ROADMAP}
+
+<files_to_read>
+- {PHASE_DIR}/*-SPRINT.md (Plans to verify)
+- {roadmap_path} (Roadmap)
+- {requirements_path} (Requirements)
+- {context_path} (USER DECISIONS from /rcode-discuss-phase)
+- {research_path} (Technical Research — includes Validation Architecture)
+</files_to_read>
+
+${AGENT_SKILLS_CHECKER}
+
+**Phase requirement IDs (MUST ALL be covered):** {phase_req_ids}
+
+**Project instructions:** Read ./CLAUDE.md if exists — verify plans honor project guidelines
+**Project skills:** Check .claude/skills/ or .agents/skills/ directory (if either exists) — verify plans account for project skill rules
+</verification_context>
+
+<expected_output>
+- ## VERIFICATION PASSED — all checks pass
+- ## ISSUES FOUND — structured issue list
+</expected_output>
+```
+
+```
+Task(
+  prompt=checker_prompt,
+  subagent_type="rcode-sprint-checker",
+  model="{checker_model}",
+  description="Verify Phase {phase} plans"
+)
+```
+
+## 11. Handle Checker Return
+
+- **`## VERIFICATION PASSED`:** Display confirmation, proceed to step 13.
+- **`## ISSUES FOUND`:** Display issues, check iteration count, proceed to step 12.
+
+**Thinking partner for architectural tradeoffs (conditional):**
+If `features.thinking_partner` is enabled, scan the checker's issues for architectural tradeoff keywords
+("architecture", "approach", "strategy", "pattern", "vs", "alternative"). If found:
+
+```
+The sprint-checker flagged an architectural decision point:
+{issue description}
+
+Brief analysis:
+- Option A: {approach_from_plan} — {pros/cons}
+- Option B: {alternative_approach} — {pros/cons}
+- Recommendation: {choice} aligned with {phase_goal}
+
+Apply this to the revision? [Yes] / [No, I'll decide]
+```
+
+If yes: include the recommendation in the revision prompt. If no: proceed to revision loop as normal.
+If thinking_partner disabled: skip this block entirely.
+
+## 12. Revision Loop (Max 3 Iterations, 1 in autonomous/yolo mode)
+
+**Mode-based iteration cap (token cost protection — closes #585):**
+
+```bash
+MAX_ITERATIONS=$($TOOL config-get workflow.max_checker_iterations 2>/dev/null || echo "")
+if [ -z "$MAX_ITERATIONS" ]; then
+  # Default: 1 in yolo/autonomous, 3 in guided
+  [ "$MODE" = "yolo" ] || [ -n "$AUTONOMOUS" ] && MAX_ITERATIONS=1 || MAX_ITERATIONS=3
+fi
+```
+
+Track `iteration_count` (starts at 1 after initial plan + check).
+Track `prev_issue_count` (initialized to `Infinity` before the loop begins).
+Track `stall_reentry_count` (starts at 0; incremented each time "Adjust approach" re-enters step 8).
+
+**If iteration_count < MAX_ITERATIONS:**
+
+**Sprint-checker malfunction guard (BLOCKER-class — added in v3.1.0 after #440):**
+
+Before parsing issues, verify the checker actually invoked tools. The checker MUST exhibit at least one of these evidence markers in its return:
+
+- A YAML `issues:` block (even an empty one — `issues: []`)
+- A YAML `verified_files:` block listing files it read
+- At least one `path:` field in any block (e.g. `path: src/components/Foo.tsx:42`)
+- A summary line of the form `Verified N of M files` or `Checked N symbols`
+
+If NONE of these evidence markers are present, the checker malfunctioned (returned narrative without invoking tools — see #440). BLOCK execution:
+
+```
+Display: "Sprint-checker returned without evidence of tool use — likely
+         malfunctioned (cf. issue #440). Refusing to advance the plan
+         on unverified output. Re-run /rcode-plan or inspect the agent."
+Halt the workflow with a non-zero exit signal.
+```
+
+Do NOT treat empty / narrative-only checker output as "plan approved". An empty checker output is a malfunction, not a pass.
+
+Parse issue count from checker return: count BLOCKER + WARNING entries in the YAML issues block (structured output from rcode-sprint-checker). If the checker's return contains a populated YAML issues block with `issues: []` (i.e., the plan was approved with no issues AFTER actual checking), treat `issue_count` as 0 and skip the stall check — the plan passed. Proceed to step 13.
+
+Display: `Revision iteration {N}/3 -- {blocker_count} blockers, {warning_count} warnings`
+
+**Stall detection:** If `issue_count >= prev_issue_count`:
+  Display: `Revision loop stalled — issue count not decreasing ({issue_count} issues remain after {N} iterations)`
+
+  **If `stall_reentry_count < 2`:**
+    Ask user:
+      Question: "Issues remain after {N} revision attempts with no progress. Proceed with current output?"
+      Options: "Proceed anyway" | "Adjust approach"
+    If "Proceed anyway": accept current plans and continue to step 13.
+    If "Adjust approach": increment `stall_reentry_count`, open freeform discussion, then re-enter step 8 (full replanning). Note: re-entry resets `iteration_count` and `prev_issue_count` but `stall_reentry_count` persists across re-entries and is capped at 2.
+
+  **If `stall_reentry_count >= 2`:**
+    Display: `Stall persists after 2 re-planning attempts. The following issues could not be resolved automatically:`
+    List the remaining issues from the checker.
+    Suggest: "Consider resolving these issues manually or running `/rcode-debug` to investigate root causes."
+    Options: "Proceed anyway" | "Abandon"
+    If "Proceed anyway": accept current plans and continue to step 13.
+    If "Abandon": stop workflow.
+
+Set `prev_issue_count = issue_count`.
+
+Revision prompt:
+
+```markdown
+<revision_context>
+**Phase:** {phase_number}
+**Mode:** revision
+
+<files_to_read>
+- {PHASE_DIR}/*-SPRINT.md (Existing plans)
+- {context_path} (USER DECISIONS from /rcode-discuss-phase)
+</files_to_read>
+
+${AGENT_SKILLS_PLANNER}
+
+**Checker issues:** {structured_issues_from_checker}
+</revision_context>
+
+<instructions>
+Make targeted updates to address checker issues.
+Do NOT replan from scratch unless issues are fundamental.
+Return what changed.
+</instructions>
+```
+
+```
+Task(
+  prompt=revision_prompt,
+  subagent_type="rcode-planner",
+  model="{planner_model}",
+  description="Revise Phase {phase} plans"
+)
+```
+
+After planner returns -> spawn checker again (step 10), increment iteration_count.
+
+**If iteration_count >= 3:**
+
+Display: `Max iterations reached. {N} issues remain:` + issue list
+
+Offer: 1) Force proceed, 2) Provide guidance and retry, 3) Abandon
+
+## 12.5. Wave Parallelism File-Overlap Check (added in v3.1.0 after #442)
+
+Before declaring plans ready, validate the wave-parallelism rule the planner declares: **same wave + overlapping `files_modified` = sequential, not parallel**. If two plans share `depends_on` (same wave) and both list the same file in `files_modified`, the planner should have marked the later one `sequential: true`. Catch the cases where it didn't.
+
+```bash
+# For every pair of plans (A, B) with the same depends_on:
+#   if files_modified(A) ∩ files_modified(B) is non-empty:
+#     - the later plan (by sprint id) MUST declare sequential: true
+#     - and must list the conflicting files in its frontmatter
+
+node ".rcode/bin/rcode-tools.cjs" plan check-wave-overlaps "${PHASE_NUMBER}"
+```
+
+The CLI helper returns a JSON report:
+
+```json
+{
+  "conflicts": [
+    {
+      "wave": 2,
+      "plan_a": "96.2",
+      "plan_b": "96.3",
+      "shared_files": ["src/components/LeadDetailPanel.tsx", "src/styles/inbox.css"],
+      "plan_b_sequential": false
+    }
+  ]
+}
+```
+
+**If `conflicts` is non-empty:**
+
+1. For each conflict, edit the later plan's SPRINT.md frontmatter to add:
+   ```yaml
+   sequential: true
+   sequential_after: <plan_a id>
+   conflicting_files: [<shared_files...>]
+   ```
+2. Recompute waves: the formerly-parallel plan now depends on the earlier one, so its wave is `max(waves of dependencies) + 1`.
+3. Re-run the checker to confirm the updated frontmatter.
+4. Display: `Wave parallelism: {N} conflict(s) auto-corrected to sequential.`
+
+**If `conflicts` is empty:** Display `Wave parallelism: ✓ no file-overlap conflicts.` and proceed.
+
+This closes the gap from #442 — the rule was stated in `rcode-planner.md` but not enforced. Now it's enforced automatically.
+
+## 13. Requirements Coverage Gate
+
+After plans pass the checker (or checker is skipped), verify that all phase requirements are covered by at least one plan.
+
+**Skip if:** `phase_req_ids` is null or TBD (no requirements mapped to this phase).
+
+**Step 1: Extract requirement IDs claimed by plans**
+```bash
+# Collect all requirement IDs from plan frontmatter
+PLAN_REQS=$(grep -h "requirements_addressed\|requirements:" ${PHASE_DIR}/*-SPRINT.md 2>/dev/null | tr -d '[]' | tr ',' '\n' | sed 's/^[[:space:]]*//' | sort -u)
+```
+
+**Step 2: Compare against phase requirements from ROADMAP**
+
+For each REQ-ID in `phase_req_ids`:
+- If REQ-ID appears in `PLAN_REQS` → covered ✓
+- If REQ-ID does NOT appear in any plan → uncovered ✗
+
+**Step 3: Check CONTEXT.md features against plan objectives**
+
+Read CONTEXT.md `<decisions>` section. Extract feature/capability names. Check each against plan `<objective>` blocks. Features not mentioned in any plan objective → potentially dropped.
+
+**Step 4: Report**
+
+If all requirements covered and no dropped features:
+```
+✓ Requirements coverage: {N}/{N} REQ-IDs covered by plans
+```
+→ Proceed to step 14.
+
+If gaps found:
+```
+## ⚠ Requirements Coverage Gap
+
+{M} of {N} phase requirements are not assigned to any plan:
+
+| REQ-ID | Description | Plans |
+|--------|-------------|-------|
+| {id} | {from REQUIREMENTS.md} | None |
+
+{K} CONTEXT.md features not found in plan objectives:
+- {feature_name} — described in CONTEXT.md but no plan covers it
+
+Options:
+1. Re-plan to include missing requirements (recommended)
+2. Move uncovered requirements to next phase
+3. Proceed anyway — accept coverage gaps
+```
+
+If `TEXT_MODE` is true, present as a plain-text numbered list (options already shown in the block above). Otherwise use AskUserQuestion to present the options.
+
+## 13b. Record Planning Completion in STATE.md
+
+After plans pass all gates, record that planning is complete so STATE.md reflects the new phase status:
+
+```bash
+node ".rcode/bin/rcode-tools.cjs" state planned-phase --phase "${PHASE_NUMBER}" --name "${PHASE_NAME}" --plans "${PLAN_COUNT}"
+```
+
+This updates STATUS to "Ready to execute", sets the correct plan count, and timestamps Last Activity.
+
+## 14. Present Final Status
+
+Route to `<offer_next>` OR `auto_advance` depending on flags/config.
+
+## 15. Auto-Advance Check
+
+Check for auto-advance trigger:
+
+1. Parse `--auto` and `--chain` flags from $ARGUMENTS
+2. **Sync chain flag with intent** — if user invoked manually (no `--auto` and no `--chain`), clear the ephemeral chain flag from any previous interrupted `--auto` chain. This does NOT touch `workflow.auto_advance` (the user's persistent settings preference):
+   ```bash
+   if [[ ! "$ARGUMENTS" =~ --auto ]] && [[ ! "$ARGUMENTS" =~ --chain ]]; then
+     node ".rcode/bin/rcode-tools.cjs" config-set workflow._auto_chain_active false 2>/dev/null
+   fi
+   ```
+3. Read both the chain flag and user preference:
+   ```bash
+   AUTO_CHAIN=$(node ".rcode/bin/rcode-tools.cjs" config-get workflow._auto_chain_active 2>/dev/null || echo "false")
+   AUTO_CFG=$(node ".rcode/bin/rcode-tools.cjs" config-get workflow.auto_advance 2>/dev/null || echo "false")
+   ```
+
+**If `--auto` or `--chain` flag present AND `AUTO_CHAIN` is not true:** Persist chain flag to config (handles direct invocation without prior discuss-phase):
+```bash
+if ([[ "$ARGUMENTS" =~ --auto ]] || [[ "$ARGUMENTS" =~ --chain ]]) && [[ "$AUTO_CHAIN" != "true" ]]; then
+  node ".rcode/bin/rcode-tools.cjs" config-set workflow._auto_chain_active true
+fi
+```
+
+**If `--auto` or `--chain` flag present OR `AUTO_CHAIN` is true OR `AUTO_CFG` is true:**
+
+Display banner:
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTO-ADVANCING TO EXECUTE
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Plans ready. Launching execute-phase...
+```
+
+Launch execute-phase using the Skill tool to avoid nested Task sessions (which cause runtime freezes due to deep agent nesting):
+```
+Skill(skill="rcode-execute", args="${PHASE} --auto --no-transition ${RCODE_WS}")
+```
+
+The `--no-transition` flag tells execute-phase to return status after verification instead of chaining further. This keeps the auto-advance chain flat — each phase runs at the same nesting level rather than spawning deeper Task agents.
+
+**Handle execute-phase return:**
+- **PHASE COMPLETE** → Display final summary:
+  ```
+  ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+   rcode ► PHASE ${PHASE} COMPLETE ✓
+  ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+  Auto-advance pipeline finished.
+
+  Next: /rcode-discuss-phase ${NEXT_PHASE} --auto ${RCODE_WS}
+  ```
+- **GAPS FOUND / VERIFICATION FAILED** → Display result, stop chain:
+  ```
+  Auto-advance stopped: Execution needs review.
+
+  Review the output above and continue manually:
+  /rcode-execute ${PHASE} ${RCODE_WS}
+  ```
+
+**If neither `--auto` nor config enabled:**
+Route to `<offer_next>` (existing behavior).
+
+</process>
+
+<banner_emission_gate>
+Issue #655 — the success banner is gated on real verification, not vibes.
+Before emitting `PLANNED ✓`, confirm one of these is true:
+
+1. A passing CHECK.md exists at `${PHASE_DIR}/*-CHECK.md` from rcode-sprint-checker
+   in this run AND its overall verdict is `pass` (or `pass-with-cautions`).
+2. The user has explicitly said "skip verification" / "override" this run AND that
+   override is recorded in the offer-next output's `Verification:` field as
+   `Passed with override`.
+3. `plan_checker_enabled` is false in config — recorded as `Verification: Skipped
+   (config-disabled)`.
+
+If none of the three holds (sprint-checker was never spawned, or it returned a
+fail verdict, or its CHECK.md is missing) — DO NOT emit `PLANNED ✓`. Emit:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► PHASE {X} PLANNED ⚠ (gates skipped)
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Plans were written but rcode-sprint-checker did not return a passing
+CHECK.md. Run /rcode-plan {X} --reviews to gate the plans before
+executing, or pass --skip-verify if you accept the risk.
+```
+
+The same rule applies to `VERIFIED ✓` (after /rcode-verify-phase) and
+`DONE ✓` (after /rcode-execute) — the success-tick is reserved for
+gate-passed states.
+</banner_emission_gate>
+
+<offer_next>
+Output this markdown directly (not as a code block):
+
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► PHASE {X} PLANNED ✓
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+**Phase {X}: {Name}** — {N} plan(s) in {M} wave(s)
+
+| Wave | Plans | What it builds |
+|------|-------|----------------|
+| 1    | 01, 02 | [objectives] |
+| 2    | 03     | [objective]  |
+
+Research: {Completed | Used existing | Skipped}
+Verification: {Passed | Passed with override | Skipped}
+
+───────────────────────────────────────────────────────────────
+
+## ▶ Next Up
+
+**Execute Phase {X}** — run all {N} plans
+
+/clear then:
+
+/rcode-execute {X} ${RCODE_WS}
+
+───────────────────────────────────────────────────────────────
+
+**Also available:**
+- cat .planning/phases/{phase-dir}/*-SPRINT.md — review plans
+- /rcode-plan {X} --research — re-research first
+- /rcode-review --phase {X} --all — peer review plans with external AIs
+- /rcode-plan {X} --reviews — replan incorporating review feedback
+
+───────────────────────────────────────────────────────────────
+</offer_next>
+
+<windows_troubleshooting>
+**Windows users:** If sprint-plan freezes during agent spawning (common on Windows due to
+stdio deadlocks with MCP servers — see Claude Code issue anthropics/claude-code#28126):
+
+1. **Force-kill:** Close the terminal (Ctrl+C may not work)
+2. **Clean up orphaned processes:**
+   ```powershell
+   # Kill orphaned node processes from stale MCP servers
+   Get-Process node -ErrorAction SilentlyContinue | Where-Object {$_.StartTime -lt (Get-Date).AddHours(-1)} | Stop-Process -Force
+   ```
+3. **Clean up stale task directories:**
+   ```powershell
+   # Remove stale subagent task dirs (Claude Code never cleans these on crash)
+   Remove-Item -Recurse -Force "$env:USERPROFILE\.claude\tasks\*" -ErrorAction SilentlyContinue
+   ```
+4. **Reduce MCP server count:** Temporarily disable non-essential MCP servers in settings.json
+5. **Retry:** Restart Claude Code and run `/rcode-plan` again
+
+If freezes persist, try `--skip-research` to reduce the agent chain from 3 to 2 agents:
+```
+/rcode-plan N --skip-research
+```
+</windows_troubleshooting>
+
+<success_criteria>
+- [ ] .planning/ directory validated
+- [ ] Phase validated against roadmap
+- [ ] Phase directory created if needed
+- [ ] CONTEXT.md loaded early (step 4) and passed to ALL agents
+- [ ] Research completed (unless --skip-research or --gaps or exists)
+- [ ] rcode-phase-researcher spawned with CONTEXT.md
+- [ ] Existing plans checked
+- [ ] rcode-planner spawned with CONTEXT.md + RESEARCH.md
+- [ ] Plans created (PLANNING COMPLETE or CHECKPOINT handled)
+- [ ] rcode-sprint-checker spawned with CONTEXT.md
+- [ ] Verification passed OR user override OR max iterations with user decision
+- [ ] User sees status between agent spawns
+- [ ] User knows next steps
+</success_criteria>
+
+## Next Up
+
+- `/rcode-execute` — execute the SPRINT.md plans the planner produced
+- `/rcode-discuss-phase` — revisit decisions if the sprint-checker flagged grey areas
+- `/rcode-research-phase` — run deeper research if RESEARCH.md was skipped