@hanzlaa/rcode 3.6.15 → 4.0.0

package/rcode/workflows/autonomous.md

@@ -0,0 +1,936 @@
+<purpose>
+
+Drive milestone phases autonomously — all remaining phases, a range via `--from N`/`--to N`, or a single phase via `--only N`. For each incomplete phase: discuss → plan → execute using Skill() flat invocations. Pauses only for explicit user decisions (grey area acceptance, blockers, validation requests). Re-reads ROADMAP.md after each phase to catch dynamically inserted phases.
+
+</purpose>
+
+<critical_rules priority="absolute">
+
+These rules apply throughout autonomous execution. Violations broke the
+interpos audit (issue #221) — DO NOT regress.
+
+1. **NEVER modify `.rcode/config.yaml`.** Specifically: never write
+   `mode: yolo`, never call `rcode-tools config-set mode`, never `sed`
+   the file. The user's mode preference is sacred. Autonomous behavior
+   is governed by the workflow's own internal flags + the `--auto`
+   invocation flag, NOT by mutating persistent config.
+
+2. **NEVER skip the methodology chain on greenfield projects.** Before
+   the phase loop runs, the prerequisite check (next step) MUST verify:
+   - `.planning/prd.md` exists (else halt → /rcode-create-prd)
+   - ROADMAP.md has milestone structure (else halt → /rcode-new-milestone)
+   - `.planning/epics.md` exists (else halt → /rcode-create-epics-and-stories)
+   See issue #219 + #229.
+
+3. **NEVER write SPRINT.md directly.** Sprint creation MUST go through
+   the `rcode-sprint-planning` skill so the capacity gate (#127) fires.
+   If autonomous needs a sprint, invoke the skill — don't shortcut.
+
+4. **ALWAYS call `state sync --from-disk` after writing any
+   .planning/ artifact.** Otherwise state.json drifts and downstream
+   workflows lie. See `_shared/state-sync-rule.md` (#198).
+
+5. **ALWAYS record decisions via `rcode-tools state add-decision`.**
+   Never write decision prose to STATE.md. See #224.
+
+</critical_rules>
+
+<required_reading>
+
+@.rcode/references/output-format.md
+@.rcode/references/workstream-flag.md
+@.rcode/references/output-realism.md
+@rcode/brain/best-practices/no-autonomous-bypass.md
+@rcode/brain/best-practices/state-sync-rule.md
+@.rcode/references/karpathy-guidelines.md
+
+Read all files referenced by the invoking prompt's execution_context before starting.
+
+</required_reading>
+
+<step name="prerequisite_check" priority="before-everything">
+
+## 0. Prerequisite check (greenfield guard)
+
+Before any phase work, verify the methodology chain has run:
+
+```bash
+HAS_PRD=$( ( ls .planning/prd.md .planning/PRD.md .planning/prds/*.md .planning/milestones/*/PRD.md 2>/dev/null | head -1 ) && echo true || echo false)
+HAS_ROADMAP_MILESTONES=$(grep -qE "^## Milestone\s+M[0-9]+" .planning/ROADMAP.md 2>/dev/null && echo true || echo false)
+HAS_EPICS=$( ( ls .planning/epics.md .planning/EPICS.md .planning/epics/*.md .planning/milestones/*/EPICS.md 2>/dev/null | head -1 ) && echo true || echo false)
+SKIP_FLAG=$(echo "$ARGUMENTS" | grep -qE "\-\-skip-prerequisites" && echo true || echo false)
+```
+
+If `SKIP_FLAG=false` AND any prerequisite is missing, HALT with a clear message:
+
+```
+⚠ Cannot run autonomous: missing prerequisite — {what}.
+
+The autonomous flow assumes a project that has already gone through:
+  1. /rcode-create-prd               → produces .planning/prd.md
+  2. /rcode-new-milestone         → produces ROADMAP.md with M1..Mn
+  3. /rcode-create-epics-and-stories → produces .planning/epics.md
+  4. THEN /rcode-autonomous           ← you are here
+
+Suggested first step: /rcode-{first-missing-command}
+
+If you genuinely want to skip these (rare — usually inverted methodology),
+re-invoke with: /rcode-autonomous --skip-prerequisites
+```
+
+If `SKIP_FLAG=true`: print a warning that downstream workflows may produce low-quality output without upstream artifacts, then proceed.
+
+</step>
+
+<step name="prepare_branch">
+
+## 1b. Prepare Branch
+
+Autonomous mode MUST NOT run directly on `main` or `master` without explicit opt-in.
+This prevents accidental commits to the default branch during long autonomous runs.
+
+```bash
+CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo "unknown")
+ALLOW_MAIN=""
+if echo "$ARGUMENTS" | grep -q '\-\-allow-main'; then
+  ALLOW_MAIN="true"
+fi
+```
+
+**If `CURRENT_BRANCH` is `main` or `master` AND `ALLOW_MAIN` is NOT set:**
+
+Create a working branch:
+
+```bash
+BRANCH_NAME="rcode/autonomous-${milestone_version}-$(date +%Y%m%d-%H%M%S)"
+git checkout -b "${BRANCH_NAME}"
+```
+
+Display:
+
+```
+🔀 Created branch: ${BRANCH_NAME}
+   (autonomous mode does not run on main/master — use --allow-main to override)
+```
+
+**If `CURRENT_BRANCH` is `main` or `master` AND `ALLOW_MAIN` is set:**
+
+Display warning:
+
+```
+⚠ Running on ${CURRENT_BRANCH} — --allow-main override active
+```
+
+Proceed without branch creation.
+
+**If `CURRENT_BRANCH` is any other branch:** Proceed silently — user intentionally set up a working branch.
+
+Store `CURRENT_BRANCH` for the lifecycle step — if autonomous created the branch, it will suggest a PR at completion.
+
+</step>
+
+<step name="initialize" priority="first">
+
+## 1. Initialize
+
+Parse `$ARGUMENTS` for `--from N`, `--to N`, `--only N`, and `--interactive` flags:
+
+```bash
+FROM_PHASE=""
+if echo "$ARGUMENTS" | grep -qE '\-\-from\s+[0-9]'; then
+  FROM_PHASE=$(echo "$ARGUMENTS" | grep -oE '\-\-from\s+[0-9]+\.?[0-9]*' | awk '{print $2}')
+fi
+
+TO_PHASE=""
+if echo "$ARGUMENTS" | grep -qE '\-\-to\s+[0-9]'; then
+  TO_PHASE=$(echo "$ARGUMENTS" | grep -oE '\-\-to\s+[0-9]+\.?[0-9]*' | awk '{print $2}')
+fi
+
+ONLY_PHASE=""
+if echo "$ARGUMENTS" | grep -qE '\-\-only\s+[0-9]'; then
+  ONLY_PHASE=$(echo "$ARGUMENTS" | grep -oE '\-\-only\s+[0-9]+\.?[0-9]*' | awk '{print $2}')
+  FROM_PHASE="$ONLY_PHASE"
+fi
+
+INTERACTIVE=""
+if echo "$ARGUMENTS" | grep -q '\-\-interactive'; then
+  INTERACTIVE="true"
+fi
+```
+
+When `--only` is set, also set `FROM_PHASE` to the same value so existing filter logic applies.
+
+When `--interactive` is set, discuss runs inline with questions (not auto-answered), while plan and execute are dispatched as background agents. This keeps the main context lean — only discuss conversations accumulate — while preserving user input on all design decisions.
+
+Bootstrap via rcode-tools init + state:
+
+```bash
+INIT=$(node .rcode/bin/rcode-tools.cjs init milestone-op 2>/dev/null || node .rcode/bin/rcode-tools.cjs init)
+if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+STATE=$(node .rcode/bin/rcode-tools.cjs state read 2>/dev/null || echo '{}')
+```
+
+Parse JSON for: `milestone_version`, `milestone_name`, `phase_count`, `completed_phases`, `roadmap_exists`, `state_exists`, `commit_docs`, `response_language`.
+
+**If `response_language` is set:** include `Respond in {response_language}.` in all spawned subagent prompts.
+
+**If `roadmap_exists` is false:** Error — "No ROADMAP.md found. Run `/rcode-new-milestone` first."
+**If `state_exists` is false:** Error — "No STATE.md found. Run `/rcode-new-milestone` first."
+
+Display startup banner:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ Milestone: {milestone_version} — {milestone_name}
+ Phases: {phase_count} total, {completed_phases} complete
+```
+
+If `ONLY_PHASE` is set, display: `Single phase mode: Phase ${ONLY_PHASE}`
+Else if `FROM_PHASE` is set, display: `Starting from phase ${FROM_PHASE}`
+If `TO_PHASE` is set, display: `Stopping after phase ${TO_PHASE}`
+If `INTERACTIVE` is set, display: `Mode: Interactive (discuss inline, plan+execute in background)`
+
+</step>
+
+<step name="discover_phases">
+
+## 2. Discover Phases
+
+Parse ROADMAP.md directly (rcode-tools does not expose `roadmap analyze`):
+
+```bash
+cat .planning/ROADMAP.md
+# For per-phase detail, inspect .planning/phases/<phase_slug>/ directory for SPRINT.md, SUMMARY.md presence
+```
+
+Build an internal `phases` array with: `number`, `name`, `goal`, `disk_status` (complete if SUMMARY.md exists, partial if SPRINT.md exists without SUMMARY.md, planned if neither), `has_ui_hint`.
+
+**Filter to incomplete phases:** Keep only phases where `disk_status !== "complete"`.
+
+**Apply `--from N` filter:** If `FROM_PHASE` was provided, additionally filter out phases where `number < FROM_PHASE` (use numeric comparison — handles decimal phases like "5.1").
+
+**Apply `--to N` filter:** If `TO_PHASE` was provided, additionally filter out phases where `number > TO_PHASE` (use numeric comparison). This limits execution to phases up through the target phase.
+
+**Apply `--only N` filter:** If `ONLY_PHASE` was provided, additionally filter OUT phases where `number != ONLY_PHASE`. This means the phase list will contain exactly one phase (or zero if already complete).
+
+**If `TO_PHASE` is set and no phases remain:**
+
+```
+All phases through ${TO_PHASE} are already completed. Nothing to do.
+```
+
+Exit cleanly.
+
+**If `ONLY_PHASE` is set and no phases remain:**
+
+```
+Phase ${ONLY_PHASE} is already complete. Nothing to do.
+```
+
+Exit cleanly.
+
+**Sort by `number`** in numeric ascending order.
+
+**If no incomplete phases remain:**
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ COMPLETE 🎉
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ All phases complete! Nothing left to do.
+```
+
+Exit cleanly.
+
+**Display phase plan:**
+
+```
+## Phase Plan
+
+| # | Phase | Status |
+|---|-------|--------|
+| 5 | Skill Scaffolding & Phase Discovery | In Progress |
+| 6 | Smart Discuss | Not Started |
+| 7 | Auto-Chain Refinements | Not Started |
+| 8 | Lifecycle Orchestration | Not Started |
+```
+
+For each phase, extract `phase_name`, `goal`, `success_criteria` by reading its section in ROADMAP.md. Store for use in execute_phase and transition messages.
+
+Use TaskCreate to register a task per incomplete phase.
+
+</step>
+
+<step name="execute_phase">
+
+## 3. Execute Phase
+
+**Before displaying the banner, re-derive T from disk** (compaction guard — same
+logic as iterate step 4.0):
+
+```bash
+PROGRESS_REFRESH=$(node .rcode/bin/rcode-tools.cjs progress init 2>/dev/null)
+```
+
+Update `phase_count` and `completed_phases` from the refresh. This ensures
+the banner always shows the correct total, even after context compaction.
+
+For the current phase, display the progress banner:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ Phase {N}/{T}: {Name} [████░░░░] {P}%
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
+
+Where N = current phase number (from the ROADMAP, e.g., 63), T = total milestone phases (from `phase_count` parsed in initialize step, e.g., 67). **Important:** T must be `phase_count` (the total number of phases in this milestone), NOT the count of remaining/incomplete phases. P = percentage of all milestone phases completed so far — (number of phases with SUMMARY.md / T × 100). Use █ for filled and ░ for empty segments in the progress bar (8 characters wide).
+
+**Alternative display when phase numbers exceed total** (multi-milestone projects where phases are numbered globally): If N > T, use the format `Phase {N} ({position}/{T})` where `position` is the 1-based index among incomplete phases being processed. This prevents confusing displays like "Phase 63/5".
+
+### 3a. Smart Discuss
+
+Check if CONTEXT.md already exists for this phase:
+
+```bash
+PHASE_SLUG="<zero-padded-phase-number>-<phase-slug>"
+PHASE_DIR=".planning/phases/${PHASE_SLUG}"
+HAS_CONTEXT=$([ -f "${PHASE_DIR}/${PADDED_PHASE}-CONTEXT.md" ] || [ -f "${PHASE_DIR}/CONTEXT.md" ] && echo true || echo false)
+```
+
+**If has_context is true:** Skip discuss — context already gathered. Display:
+
+```
+Phase ${PHASE_NUMBER}: Context exists — skipping discuss.
+```
+
+Proceed to 3b.
+
+**If has_context is false:** Check if discuss is disabled via settings:
+
+```bash
+SKIP_DISCUSS=$(node .rcode/bin/rcode-tools.cjs config 2>/dev/null | grep -oE '"skip_discuss"[^,}]*' | grep -oE 'true|false' || echo "false")
+```
+
+**If SKIP_DISCUSS is `true`:** Skip discuss entirely — the ROADMAP phase description is the spec. Display:
+
+```
+Phase ${PHASE_NUMBER}: Discuss skipped (workflow.skip_discuss=true) — using ROADMAP phase goal as spec.
+```
+
+Write a minimal CONTEXT.md so downstream plan-phase has valid input. Extract `goal` and `requirements` from ROADMAP.md for this phase. Write `${PHASE_DIR}/${PADDED_PHASE}-CONTEXT.md` with:
+
+```markdown
+# Phase {PHASE_NUMBER}: {Phase Name} - Context
+
+**Gathered:** {date}
+**Status:** Ready for planning
+**Mode:** Auto-generated (discuss skipped via workflow.skip_discuss)
+
+<domain>
+## Phase Boundary
+
+{goal from ROADMAP phase description}
+
+</domain>
+
+<decisions>
+## Implementation Decisions
+
+### Claude's Discretion
+All implementation choices are at Claude's discretion — discuss phase was skipped per user setting. Use ROADMAP phase goal, success criteria, and codebase conventions to guide decisions.
+
+</decisions>
+
+<canonical_refs>
+## Canonical References
+
+No external specs pre-gathered — discuss phase skipped. Planner will extract canonical refs from ROADMAP.md during research.
+
+</canonical_refs>
+
+<code_context>
+## Existing Code Insights
+
+Codebase context will be gathered during plan-phase research.
+
+</code_context>
+
+<specifics>
+## Specific Ideas
+
+No specific requirements — discuss phase skipped. Refer to ROADMAP phase description and success criteria.
+
+</specifics>
+
+<deferred>
+## Deferred Ideas
+
+None — discuss phase skipped.
+
+</deferred>
+```
+
+Commit the minimal context (guarded for gitignored `.planning/`):
+
+```bash
+git add "${PHASE_DIR}/${PADDED_PHASE}-CONTEXT.md" 2>/dev/null \
+  && git commit -m "docs(${PADDED_PHASE}): auto-generated context (discuss skipped)" 2>/dev/null \
+  || echo "ℹ .planning/ gitignored — context written, not committed"
+```
+
+Proceed to 3b.
+
+**If SKIP_DISCUSS is `false` (or unset):**
+
+**IMPORTANT — Discuss must be single-pass in autonomous mode.**
+The discuss step in autonomous mode MUST NOT loop. If CONTEXT.md already exists after discuss completes, do NOT re-invoke discuss for the same phase. The has_context check below is authoritative.
+
+**If `INTERACTIVE` is set:** Run the standard discuss-phase skill inline (asks interactive questions, waits for user answers):
+
+```
+Skill(skill="rcode-discuss-phase", args="${PHASE_NUMBER}")
+```
+
+**If `INTERACTIVE` is NOT set:** Execute the smart_discuss step for this phase (batch table proposals, auto-optimized — see smart_discuss step below).
+
+After discuss completes (either mode), verify context was written by checking for CONTEXT.md. If not present → go to handle_blocker: "Discuss for phase ${PHASE_NUMBER} did not produce CONTEXT.md."
+
+### 3a.5. UI Design Contract (Frontend Phases)
+
+Check if this phase has frontend indicators and whether a UI-SPEC already exists:
+
+```bash
+PHASE_SECTION=$(sed -n "/^## Phase ${PHASE_NUMBER}/,/^## Phase /p" .planning/ROADMAP.md)
+echo "$PHASE_SECTION" | grep -iE "UI|interface|frontend|component|layout|page|screen|view|form|dashboard|widget" > /dev/null 2>&1
+HAS_UI=$?
+UI_SPEC_FILE=$(ls "${PHASE_DIR}"/*-UI-SPEC.md 2>/dev/null | head -1)
+UI_PHASE_CFG=$(node .rcode/bin/rcode-tools.cjs config 2>/dev/null | grep -oE '"ui_phase"[^,}]*' | grep -oE 'true|false' || echo "true")
+```
+
+**If `HAS_UI` is 0 (frontend indicators found) AND `UI_SPEC_FILE` is empty AND `UI_PHASE_CFG` is not `false`:**
+
+Display:
+
+```
+Phase ${PHASE_NUMBER}: Frontend phase detected — generating UI design contract...
+```
+
+```
+Skill(skill="rcode-ui-phase", args="${PHASE_NUMBER}")
+```
+
+Verify UI-SPEC was created. If still empty after ui-phase, display a non-blocking warning and proceed to 3b.
+
+**Otherwise:** Skip silently to 3b.
+
+### 3b. Plan
+
+**If `INTERACTIVE` is set:** Dispatch plan as a background Task agent to keep the main context lean:
+
+```
+Task(
+  description="Plan phase ${PHASE_NUMBER}: ${PHASE_NAME}",
+  subagent_type="rcode-planner",
+  run_in_background=true,
+  prompt="Run plan-phase for phase ${PHASE_NUMBER}: Skill(skill=\"rcode-plan\", args=\"${PHASE_NUMBER}\")"
+)
+```
+
+Store the agent task_id. After discuss for the next phase completes (or if no next phase), wait for the plan agent to finish before proceeding to execute.
+
+**If `INTERACTIVE` is NOT set (default):** Run plan inline as before.
+
+```
+Skill(skill="rcode-plan", args="${PHASE_NUMBER}")
+```
+
+Verify plan produced output — check `${PHASE_DIR}` for `*-PLAN.md` or `SPRINT.md`. If none → go to handle_blocker: "Plan phase ${PHASE_NUMBER} did not produce any plans."
+
+### 3c. Execute
+
+**If `INTERACTIVE` is set:** Wait for the plan agent to complete (if not already), verify plans exist, then dispatch execute as a background agent:
+
+```
+Task(
+  description="Execute phase ${PHASE_NUMBER}: ${PHASE_NAME}",
+  subagent_type="rcode-executor",
+  run_in_background=true,
+  prompt="Run execute-phase for phase ${PHASE_NUMBER}: Skill(skill=\"rcode-execute\", args=\"${PHASE_NUMBER} --no-transition\")"
+)
+```
+
+Store the agent task_id. The workflow can now start discussing the next phase while this phase executes in the background. Before starting post-execution routing for this phase, wait for the execute agent to complete.
+
+**If `INTERACTIVE` is NOT set (default):** Run execute inline as before.
+
+```
+Skill(skill="rcode-execute", args="${PHASE_NUMBER} --no-transition")
+```
+
+### 3c.5. Code Review and Fix
+
+Auto-invoke code review and fix chain. Autonomous mode chains both review and fix.
+
+**Config gate:**
+```bash
+CODE_REVIEW_ENABLED=$(node .rcode/bin/rcode-tools.cjs config 2>/dev/null | grep -oE '"code_review"[^,}]*' | grep -oE 'true|false' || echo "false")
+```
+If `"false"`: display "Code review skipped (workflow.code_review=false)" and proceed to 3d.
+
+```
+Skill(skill="rcode-review", args="${PHASE_NUMBER}")
+```
+
+Parse status from REVIEW.md frontmatter. If "clean" or "skipped": proceed to 3d. If findings found: auto-invoke:
+```
+Skill(skill="rcode-review-fix", args="${PHASE_NUMBER} --auto")
+```
+
+**Error handling:** If either Skill fails, catch the error, display as non-blocking, and proceed to 3d.
+
+### 3d. Post-Execution Routing
+
+**If `INTERACTIVE` is set:** Wait for the execute agent to complete before reading verification results.
+
+After execute returns, read the verification result:
+
+```bash
+VERIFY_STATUS=$(grep "^status:" "${PHASE_DIR}"/*-VERIFICATION.md 2>/dev/null | head -1 | cut -d: -f2 | tr -d ' ')
+```
+
+**If VERIFY_STATUS is empty** (no VERIFICATION.md or no status field):
+
+Go to handle_blocker: "Execute phase ${PHASE_NUMBER} did not produce verification results."
+
+**If `passed`:**
+
+Display:
+```
+Phase ${PHASE_NUMBER} ✓ ${PHASE_NAME} — Verification passed
+```
+
+Proceed to iterate step.
+
+**If `human_needed`:**
+
+Read the human_verification section from VERIFICATION.md to get the count and items requiring manual testing.
+
+Display the items, then ask user via AskUserQuestion:
+- **question:** "Phase ${PHASE_NUMBER} has items needing manual verification. Validate now or continue to next phase?"
+- **options:** "Validate now" / "Continue without validation"
+
+On **"Validate now"**: Present the specific items from VERIFICATION.md. After user reviews, ask:
+- **question:** "Validation result?"
+- **options:** "All good — continue" / "Found issues"
+
+On "All good — continue": Display `Phase ${PHASE_NUMBER} ✓ Human validation passed` and proceed to iterate step.
+
+On "Found issues": Go to handle_blocker with the user's reported issues as the description.
+
+On **"Continue without validation"**: Display `Phase ${PHASE_NUMBER} ⏭ Human validation deferred` and proceed to iterate step.
+
+**If `gaps_found`:**
+
+Read gap summary from VERIFICATION.md (score and missing items). Display:
+```
+⚠ Phase ${PHASE_NUMBER}: ${PHASE_NAME} — Gaps Found
+Score: {N}/{M} must-haves verified
+```
+
+**If `INTERACTIVE` is set:** Ask via AskUserQuestion:
+- **question:** "Gaps found in phase ${PHASE_NUMBER}. How to proceed?"
+- **options:** "Run gap closure" / "Continue without fixing" / "Stop autonomous mode"
+- On "Stop": go to handle_blocker
+
+**If `INTERACTIVE` is NOT set (default autonomous):** Auto-select "Run gap closure" — display `⚙ Phase ${PHASE_NUMBER}: auto-running gap closure` and proceed.
+
+Execute gap closure cycle (limit: 1 attempt):
+
+```
+Skill(skill="rcode-plan", args="${PHASE_NUMBER} --gaps")
+```
+
+Verify gap plans were created. If none → go to handle_blocker: "Gap closure planning for phase ${PHASE_NUMBER} did not produce plans."
+
+Re-execute:
+```
+Skill(skill="rcode-execute", args="${PHASE_NUMBER} --no-transition")
+```
+
+Re-read verification status. If `passed` or `human_needed`: route normally. If still `gaps_found` after this retry:
+- **If `INTERACTIVE`:** ask "Continue anyway / Stop autonomous mode"
+- **Otherwise:** display `⏭ Phase ${PHASE_NUMBER}: gaps persist after closure — continuing` and proceed to iterate step.
+
+This limits gap closure to 1 automatic retry to prevent infinite loops.
+
+### 3d.5. UI Review (Frontend Phases)
+
+> Run after any successful execution routing (passed, human_needed accepted, or gaps deferred/accepted) — before proceeding to the iterate step.
+
+```bash
+UI_SPEC_FILE=$(ls "${PHASE_DIR}"/*-UI-SPEC.md 2>/dev/null | head -1)
+UI_REVIEW_CFG=$(node .rcode/bin/rcode-tools.cjs config 2>/dev/null | grep -oE '"ui_review"[^,}]*' | grep -oE 'true|false' || echo "true")
+```
+
+**If `UI_SPEC_FILE` is not empty AND `UI_REVIEW_CFG` is not `false`:**
+
+Display:
+
+```
+Phase ${PHASE_NUMBER}: Frontend phase with UI-SPEC — running UI review audit...
+```
+
+```
+Skill(skill="rcode-ui-review", args="${PHASE_NUMBER}")
+```
+
+Display the review result summary (score from UI-REVIEW.md if produced). Continue to iterate step regardless of score — UI review is advisory, not blocking.
+
+**Otherwise:** Skip silently to iterate step.
+
+</step>
+
+@rcode/workflows/autonomous-smart-discuss.md
+
+
+<step name="iterate">
+
+## 4. Iterate
+
+### 4.0. Refresh Phase Count From Disk (MANDATORY — Compaction Guard)
+
+**This step is NON-NEGOTIABLE.** LLM context compaction loses in-memory
+variables. Re-derive `phase_count` and `completed_phases` from disk at the
+START of every iteration — never rely on values held from the initialize step.
+
+```bash
+# Re-read the authoritative progress snapshot from the CLI
+PROGRESS_REFRESH=$(node .rcode/bin/rcode-tools.cjs progress init 2>/dev/null)
+```
+
+Parse `PROGRESS_REFRESH` JSON and update:
+- `phase_count` ← `PROGRESS_REFRESH.phase_count`
+- `completed_phases` ← `PROGRESS_REFRESH.completed_count`
+- `T` (for banner display) ← `phase_count`
+
+If `PROGRESS_REFRESH` fails or is empty, fall back to direct ROADMAP.md parsing:
+
+```bash
+# Fallback: count phases directly from ROADMAP.md
+PHASE_COUNT_DISK=$(grep -cE '^\|\s*\d{1,3}' .planning/ROADMAP.md 2>/dev/null || echo "0")
+COMPLETED_DISK=$(find .planning/phases/ -name '*SUMMARY.md' 2>/dev/null | wc -l)
+```
+
+**Sanity check:** If `current_phase_number > phase_count`, this is a drift
+symptom. Log a warning and re-derive from disk:
+
+```
+⚠ Phase drift detected: current phase ${N} > total ${T}. Re-reading from ROADMAP.md.
+```
+
+**If `ONLY_PHASE` is set:** Do not iterate. Proceed directly to lifecycle step (which exits cleanly per single-phase mode).
+
+**If `TO_PHASE` is set and current phase number >= `TO_PHASE`:**
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ --to ${TO_PHASE} REACHED
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ Completed through phase ${TO_PHASE} as requested.
+ Remaining phases were not executed.
+
+ Resume with: /rcode-autonomous --from ${next_incomplete_phase}
+```
+
+Proceed directly to lifecycle step (which handles partial completion). Exit cleanly.
+
+**Otherwise:** After each phase completes, re-read ROADMAP.md to catch phases inserted mid-execution (decimal phases like 5.1):
+
+```bash
+cat .planning/ROADMAP.md
+```
+
+Re-filter incomplete phases using the same logic as discover_phases.
+
+Read STATE.md fresh:
+
+```bash
+cat .planning/STATE.md
+node .rcode/bin/rcode-tools.cjs state read
+```
+
+Check for blockers in the Blockers/Concerns section. If blockers are found, go to handle_blocker.
+
+If incomplete phases remain: proceed to next phase, loop back to execute_phase.
+
+**Token cost report and /clear offer (closes #586):**
+
+After each phase completes, display a cost summary:
+
+```
+Phase {N} ✓ {name} — {sprint_count} sprints, {revision_count} revision(s)
+```
+
+Track `PHASES_COMPLETED` (increment after each iterate). Every 3 completed phases:
+
+```
+⚠ Context growing — {PHASES_COMPLETED} phases done in this session.
+  Tip: /clear now and resume to keep remaining phases lean:
+  /rcode-autonomous --from {next_phase} --to {TO_PHASE}
+  Continue anyway? [Yes, keep going] / [I'll clear now — show resume command]
+```
+
+In `mode: yolo`: skip the offer and just print the resume command as an info line.
+
+**Interactive mode overlap:** When `INTERACTIVE` is set, the iterate step enables pipeline parallelism:
+1. After discuss completes for Phase N, dispatch plan+execute as background agents
+2. Immediately start discuss for Phase N+1 while Phase N builds
+3. Before starting plan for Phase N+1, wait for Phase N's execute agent to complete and handle its post-execution routing
+
+This means the user is always answering discuss questions (lightweight, interactive) while the heavy work runs in the background.
+
+If all phases complete, proceed to lifecycle step.
+
+</step>
+
+<step name="lifecycle">
+
+## 5. Lifecycle
+
+**If `ONLY_PHASE` is set:** Skip lifecycle. Display:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ PHASE ${ONLY_PHASE} COMPLETE ✓
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ Phase ${ONLY_PHASE}: ${PHASE_NAME} — Done
+ Mode: Single phase (--only)
+
+ Lifecycle skipped — run /rcode-autonomous without --only
+ after all phases complete to trigger audit/complete/cleanup.
+```
+
+Exit cleanly.
+
+**Otherwise:** After all phases complete, run the milestone lifecycle sequence: audit → complete → cleanup.
+
+Display lifecycle transition banner:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ LIFECYCLE
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ All phases complete → Starting lifecycle: audit → complete → cleanup
+ Milestone: {milestone_version} — {milestone_name}
+```
+
+### 5a. Audit
+
+```
+Skill(skill="rcode-audit-milestone")
+```
+
+After audit completes, detect the result:
+
+```bash
+AUDIT_FILE=".planning/v${milestone_version}-MILESTONE-AUDIT.md"
+[ -f "$AUDIT_FILE" ] || AUDIT_FILE=".planning/MILESTONE-AUDIT.md"
+AUDIT_STATUS=$(grep "^status:" "${AUDIT_FILE}" 2>/dev/null | head -1 | cut -d: -f2 | tr -d ' ')
+```
+
+**If AUDIT_STATUS is empty:** Go to handle_blocker: "Audit did not produce results — audit file missing or malformed."
+
+**If `passed`:**
+
+```
+Audit ✓ passed — proceeding to complete milestone
+```
+
+Proceed to 5b.
+
+**If `gaps_found`:**
+
+Read the gaps summary from the audit file. Display:
+```
+⚠ Audit: Gaps Found
+```
+
+**If `INTERACTIVE`:** Ask via AskUserQuestion:
+- **question:** "Milestone audit found gaps. How to proceed?"
+- **options:** "Continue anyway — accept gaps" / "Stop — fix gaps manually"
+- On "Stop": Go to handle_blocker.
+
+**Otherwise (autonomous):** Display `Audit ⏭ Gaps accepted — continuing` and proceed to 5b.
+
+**If `tech_debt`:**
+
+Show the summary, then:
+
+**If `INTERACTIVE`:** Ask via AskUserQuestion:
+- **options:** "Continue with tech debt" / "Stop — address debt first"
+- On "Stop": Go to handle_blocker.
+
+**Otherwise (autonomous):** Display `Tech debt noted — continuing` and proceed to 5b.
+
+### 5b. Complete Milestone
+
+```
+Skill(skill="rcode-complete-milestone", args="${milestone_version}")
+```
+
+After complete-milestone returns, verify archive output:
+
+```bash
+ls .planning/milestones/v${milestone_version}-ROADMAP.md 2>/dev/null || true
+```
+
+If the archive file does not exist, go to handle_blocker: "Complete milestone did not produce expected archive files."
+
+### 5c. Cleanup
+
+```
+Skill(skill="rcode-cleanup")
+```
+
+Cleanup shows its own dry-run and asks user for approval internally — this is an acceptable pause since it's an explicit decision about file deletion.
+
+### 5d. Final Completion
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ COMPLETE 🎉
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ Milestone: {milestone_version} — {milestone_name}
+ Status: Complete ✓
+ Lifecycle: audit ✓ → complete ✓ → cleanup ✓
+
+ Ship it! 🚀
+```
+
+**If autonomous created a branch** (i.e., `BRANCH_NAME` was set in the prepare_branch step):
+
+Display:
+
+```
+🔀 Branch: ${BRANCH_NAME}
+   All work is on this branch. When ready, merge to main:
+     git checkout main && git merge ${BRANCH_NAME}
+   Or create a PR for review.
+```
+
+</step>
+
+<step name="handle_blocker">
+
+## 6. Handle Blocker
+
+When any phase operation fails or a blocker is detected, present 3 options via AskUserQuestion:
+
+**Prompt:** "Phase {N} ({Name}) encountered an issue: {description}"
+
+**Options:**
+1. **"Fix and retry"** — Re-run the failed step (discuss, plan, or execute) for this phase
+2. **"Skip this phase"** — Mark phase as skipped, continue to the next incomplete phase
+3. **"Stop autonomous mode"** — Display summary of progress so far and exit cleanly
+
+**On "Fix and retry":** Loop back to the failed step. If the same step fails again after retry, re-present these options.
+
+**On "Skip this phase":** Log `Phase {N} ⏭ {Name} — Skipped by user`. Record in state:
+
+```bash
+node .rcode/bin/rcode-tools.cjs state add-decision "Skipped phase ${PHASE_NUMBER} in autonomous mode"
+```
+
+Proceed to iterate.
+
+**On "Stop autonomous mode":**
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ rcode ► AUTONOMOUS ▸ STOPPED
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+ Completed: {list of completed phases}
+ Skipped: {list of skipped phases}
+ Remaining: {list of remaining phases}
+
+ Resume with: /rcode-autonomous ${ONLY_PHASE ? "--only " + ONLY_PHASE : "--from " + next_phase}${TO_PHASE ? " --to " + TO_PHASE : ""}
+```
+
+Record blocker in state:
+
+```bash
+node .rcode/bin/rcode-tools.cjs state add-blocker "Autonomous mode stopped at phase ${PHASE_NUMBER}: ${DESCRIPTION}"
+```
+
+</step>
+
+</process>
+
+<success_criteria>
+- [ ] All incomplete phases executed in order (smart discuss → ui-phase → plan → execute → ui-review each)
+- [ ] Smart discuss proposes grey area answers in tables, user accepts or overrides per area
+- [ ] Progress banners displayed between phases
+- [ ] Execute invoked with --no-transition (autonomous manages transitions)
+- [ ] Post-execution verification reads VERIFICATION.md and routes on status
+- [ ] Passed verification → automatic continue to next phase
+- [ ] Human-needed verification → user prompted to validate or skip
+- [ ] Gaps-found → user offered gap closure, continue, or stop
+- [ ] Gap closure limited to 1 retry (prevents infinite loops)
+- [ ] Plan and execute failures route to handle_blocker
+- [ ] ROADMAP.md re-read after each phase (catches inserted phases)
+- [ ] STATE.md checked for blockers before each phase
+- [ ] Blockers handled via user choice (retry / skip / stop)
+- [ ] Final completion or stop summary displayed
+- [ ] After all phases complete, lifecycle step is invoked (not manual suggestion)
+- [ ] Lifecycle transition banner displayed before audit
+- [ ] Audit invoked via Skill(skill="rcode-audit-milestone")
+- [ ] Audit result routing: passed → auto-continue, gaps_found → user decides, tech_debt → user decides
+- [ ] Complete-milestone invoked via Skill() with ${milestone_version} arg
+- [ ] Cleanup invoked via Skill() — internal confirmation is acceptable
+- [ ] Final completion banner displayed after lifecycle
+- [ ] Progress bar uses phase number / total milestone phases, with fallback when phase numbers exceed total
+- [ ] Frontend phases get UI-SPEC generated before planning (step 3a.5) if not already present
+- [ ] Frontend phases get UI review audit after successful execution (step 3d.5) if UI-SPEC exists
+- [ ] UI phase and UI review respect workflow.ui_phase and workflow.ui_review config toggles
+- [ ] UI review is advisory (non-blocking)
+- [ ] `--only N` restricts execution to exactly one phase
+- [ ] `--only N` skips lifecycle step
+- [ ] `--only N` exits cleanly after single phase completes
+- [ ] `--only N` on already-complete phase exits with message
+- [ ] `--to N` stops execution after phase N completes
+- [ ] `--to N` filters out phases with number > N during discovery
+- [ ] `--to N` displays "Stopping after phase N" in startup banner
+- [ ] `--to N` on already completed target exits with "already completed" message
+- [ ] `--to N` compatible with `--from N`
+- [ ] `--to N` skips lifecycle when not all milestone phases complete
+- [ ] `--interactive` runs discuss inline (asks questions, waits for user)
+- [ ] `--interactive` dispatches plan and execute as background agents
+- [ ] `--interactive` enables pipeline parallelism: discuss Phase N+1 while Phase N builds
+- [ ] `--interactive` main context only accumulates discuss conversations
+- [ ] `--interactive` waits for background agents before post-execution routing
+- [ ] `--interactive` compatible with `--only`, `--from`, and `--to` flags
+- [ ] No `git push` issued by the workflow (per AGENTS.md)
+- [ ] Branch created when on main/master (unless --allow-main override)
+- [ ] Branch name follows `rcode/autonomous-{version}-{timestamp}` pattern
+- [ ] --allow-main flag skips branch creation with warning
+- [ ] Non-main/master branches used as-is without branch creation
+- [ ] PR/merge suggestion displayed at lifecycle completion when branch was created
+- [ ] Phase count T re-derived from disk at start of every iteration (compaction guard)
+- [ ] Phase count T re-derived from disk before execute_phase banner display
+- [ ] Drift warning logged when current phase number exceeds total
+- [ ] Progress bar always accurate after LLM context compaction
+</success_criteria>
+
+## Next Up
+
+- `/rcode-ship` — push the branch and open a PR after autonomous run completes
+- `/rcode-complete-milestone` — mark the milestone done once all phases shipped