@pmelab/gtd 7.6.2 → 7.7.0

package/README.md

@@ -108,7 +108,9 @@ handful of commands drive it:
   elsewhere shows up without a manual refresh.
 - **`gtd edit [path]`** — open `<path>` (or, with no argument, the current
   resting state's steering file, or the repo dir) in `${VISUAL:-$EDITOR}`,
-  blocking until it exits.
+  blocking until it exits. Low-level plumbing — for "force an edit at the
+  current human gate, then keep driving", pass `--edit` to the loop instead (see
+  below).
 
 `gtd version` (or `gtd --version`/`-v`) prints the installed version and exits;
 `gtd help` (or `gtd --help`/`-h`) prints the command list. Both short-circuit
@@ -181,8 +183,12 @@ code), waits for you to save and exit, then captures your edit itself and keeps
 driving — or halts if you left nothing changed — so you never run
 `gtd step human` by hand. Pass `--no-edit` (or set `GTD_NO_EDIT`) to fall back
 to halting and printing the gate instead, if you'd rather edit and re-launch it
-yourself. Bare `gtd` prints one line per event — colored and emoji on a real
-terminal, plain ASCII under `NO_COLOR` or when piped — and redirects the noisier
+yourself; pass `--edit`/`-e` to force the editor open at the gate right now,
+overriding an ambient `GTD_NO_EDIT`/`--no-edit`. Pass `--once` to restrict a run
+to exactly one beat — one human gate, one check, or one agent turn — instead of
+driving all the way to idle; it combines freely with `--edit`/`--no-edit`. Bare
+`gtd` prints one line per event — colored and emoji on a real terminal, plain
+ASCII under `NO_COLOR` or when piped — and redirects the noisier
 agent/check/step subprocess output to a per-repo/per-worktree log file.
 `gtd log` opens that logfile in your editor. See
 [Driving the loop](docs/loop.md).

package/bin/gtd

@@ -3,21 +3,29 @@ set -euo pipefail
 
 # This file is the single published `bin` entry: `bin/gtd`. Before anything
 # else, it dispatches on $1:
-#   - `--no-edit` (bare, or immediately after `loop`) -> stripped here and
-#     disables the loop's automatic editor launching for this run (see
-#     `edit_disabled`/`GTD_NO_EDIT` below); dispatch then continues on the
-#     remaining arguments as if it had never been given.
+#   - `--no-edit` / `--edit` (`-e`) / `--once` (bare, or immediately after
+#     `loop`, in any order/combination — see the flag-parsing block below) ->
+#     stripped here; `--no-edit` disables the loop's automatic editor
+#     launching for this run, `--edit` forces it on for this run (overriding
+#     an ambient `GTD_NO_EDIT`/`--no-edit` — the two are mutually exclusive),
+#     and `--once` restricts the loop to exactly one beat (one human gate, one
+#     script check+step, or one agent prompt+step) before exiting. Dispatch
+#     then continues on the remaining arguments as if none of them had ever
+#     been given.
 #   - `edit [path]` -> handled entirely here in bash, never forwarded to the
 #     bundle: opens `${VISUAL:-$EDITOR}` on `path`, or (no `path` given) on the
-#     resolved rest's `.file` (or `.` when absent).
+#     resolved rest's `.file` (or `.` when absent). Low-level plumbing now that
+#     `--edit` covers the common "force an edit, then keep driving" case; this
+#     subcommand remains for opening a specific path, or the current rest's
+#     file, WITHOUT driving the loop afterward.
 #   - `log` -> handled entirely here in bash, never forwarded to the bundle:
 #     opens `${VISUAL:-$EDITOR}` on the current repo/worktree's loop logfile
 #     (see `resolve_log_path`), refusing rather than opening an empty buffer
 #     when no loop has produced one yet.
-#   - no arguments, or exactly `loop` with no further arguments -> fall through
-#     into the loop body below (the loop protocol comment further down).
-#   - `loop` with additional arguments (other than a lone `--no-edit`) ->
-#     usage error, exit 2.
+#   - no arguments, or exactly `loop` with no further arguments (other than the
+#     flags above) -> fall through into the loop body below (the loop protocol
+#     comment further down).
+#   - `loop` with an unrecognised extra argument -> usage error, exit 2.
 #   - anything else -> hand off to the bundle via `run_gtd`, forwarding all
 #     arguments unchanged (e.g. `bin/gtd status`, `bin/gtd step human`).
 # `run_gtd` is the ONE place that knows how to invoke "the bundle": by default
@@ -206,19 +214,59 @@ resolve_log_path() {
   fi
 }
 
-# --no-edit / GTD_NO_EDIT govern only the loop's automatic editor launching at
-# human gates (see handle_human_gate further down) — `gtd edit` itself never
-# reads either. `--no-edit` is recognised in exactly the two sanctioned
-# positions (bare, or right after `loop`) and stripped before anything below
-# looks at "$@", so the bundle's own unknown-option guard never sees it.
+# --no-edit / --edit (-e) / --once govern only the loop driver, recognised in
+# exactly two positions — bare (e.g. `gtd --edit --once`) or immediately after
+# `loop` (e.g. `gtd loop --once`) — in any order, each at most once. `--edit`
+# and `--no-edit` are mutually exclusive (forcing and suppressing the editor
+# in the same run is not a coherent request). A recognised, valid combination
+# is stripped here before anything below looks at "$@", so the bundle's own
+# unknown-option guard never sees it; anything else (an unknown flag, a
+# duplicate/conflicting one, or a flag mixed with a real subcommand) is left
+# untouched and falls through to the existing dispatch below, which forwards a
+# non-flag first argument to the bundle (whose own unknown-option guard
+# rejects it) or rejects a `loop` invocation carrying an unrecognised extra
+# argument.
 edit_disabled=0
+edit_forced=0
+once_mode=0
 [[ -n "${GTD_NO_EDIT:-}" ]] && edit_disabled=1
-if [[ "${1:-}" == "--no-edit" && $# -eq 1 ]]; then
-  edit_disabled=1
-  shift
-elif [[ "${1:-}" == "loop" && "${2:-}" == "--no-edit" && $# -eq 2 ]]; then
-  edit_disabled=1
-  set -- loop
+
+if [[ "${1:-}" == "loop" ]]; then
+  loop_form=1
+  loop_flag_args=("${@:2}")
+else
+  loop_form=0
+  loop_flag_args=("$@")
+fi
+
+valid_combo=1
+n_no_edit=0
+n_edit=0
+n_once=0
+for a in ${loop_flag_args[@]+"${loop_flag_args[@]}"}; do
+  case "$a" in
+  --no-edit) n_no_edit=$((n_no_edit + 1)) ;;
+  --edit | -e) n_edit=$((n_edit + 1)) ;;
+  --once) n_once=$((n_once + 1)) ;;
+  *) valid_combo=0 ;;
+  esac
+done
+if ((n_no_edit > 1 || n_edit > 1 || n_once > 1 || (n_no_edit >= 1 && n_edit >= 1))); then
+  valid_combo=0
+fi
+
+if [[ "$valid_combo" == 1 ]]; then
+  ((n_no_edit >= 1)) && edit_disabled=1
+  ((n_edit >= 1)) && edit_forced=1
+  ((n_once >= 1)) && once_mode=1
+  # --edit overrides an ambient GTD_NO_EDIT/--no-edit for THIS run — the
+  # explicit ask to force the editor wins over the suppressing default.
+  ((edit_forced)) && edit_disabled=0
+  if [[ "$loop_form" == 1 ]]; then
+    set -- loop
+  else
+    set --
+  fi
 fi
 
 if [[ "${1:-}" == "edit" ]]; then
@@ -382,8 +430,12 @@ Fix the reported issue and re-run gtd-loop."
     exit 0
   fi
   # A new commit was captured — render it, then return normally so the
-  # caller keeps driving.
+  # caller keeps driving, unless --once says this one human gate is the whole
+  # run.
   report_commits "$head_before"
+  if [[ "$once_mode" == 1 ]]; then
+    exit 0
+  fi
 }
 
 trap 'code=$?; if [ "$code" -ne 0 ]; then
@@ -456,7 +508,10 @@ run_agent_turn() {
 # (open the editor there first, THEN step) instead of blindly stepping human
 # and moving on. Editing OFF: unchanged from before this function existed —
 # step silently, refusal surfaced and halts, a clean/committed gate just falls
-# into the loop below.
+# into the loop below. Either way, --once treats a captured commit here as the
+# run's one beat and exits (see handle_human_gate's own once-mode exit for the
+# editing-ON path); a clean/no-op gate captures nothing, so it isn't a beat —
+# it falls into the loop below exactly as it would without --once.
 if ! peek_json="$(run_gtd next --json 2>&1)"; then
   emit_error "could not determine the next step:
 $peek_json
@@ -464,8 +519,10 @@ Run \`gtd status\` to inspect the repo."
   emit_log_pointer
   exit 1
 fi
-if [[ "$(jq -r .kind <<<"$peek_json")" == "message" ]]; then
+peek_kind="$(jq -r .kind <<<"$peek_json")"
+if [[ "$peek_kind" == "message" ]]; then
   if [[ "$edit_disabled" == 1 ]]; then
+    opening_head_before="$(git rev-parse HEAD 2>/dev/null || echo none)"
     if ! step_out="$(run_gtd step human 2>&1)"; then
       emit_error "could not capture your changes at the human gate:
 $step_out
@@ -473,9 +530,21 @@ Fix the reported issue and re-run gtd-loop."
       emit_log_pointer
       exit 1
     fi
+    if [[ "$once_mode" == 1 && "$opening_head_before" != "$(git rev-parse HEAD 2>/dev/null || echo none)" ]]; then
+      exit 0
+    fi
   else
     handle_human_gate "$peek_json"
   fi
+elif [[ "$edit_forced" == 1 ]]; then
+  # Honesty caveat: a forced edit only means something at a human gate — the
+  # machine isn't resting at one right now, so say so instead of pretending,
+  # and just keep driving.
+  if [[ "$FANCY" == 1 ]]; then
+    printf '%sℹ️  --edit: not at a human gate yet — continuing to drive%s\n' "$C_DIM" "$C_RESET"
+  else
+    printf '%s--edit: not at a human gate yet — continuing to drive%s\n' "$C_DIM" "$C_RESET"
+  fi
 fi
 
 while true; do
@@ -527,6 +596,9 @@ while true; do
     fi
     prev_state=""
     prev_content=""
+    if [[ "$once_mode" == 1 ]]; then
+      exit 0
+    fi
     continue
   fi
 
@@ -610,4 +682,7 @@ Stopping rather than stepping with a malformed steering file."
   # zero commits here — report_commits returns non-zero, and that's fine: the
   # NEXT iteration's stall check is what catches unchanged state/content.
   report_commits "$head_before" || true
+  if [[ "$once_mode" == 1 ]]; then
+    exit 0
+  fi
 done

package/dist/gtd.bundle.mjs

@@ -190594,7 +190594,7 @@ const compileWorkflowConfig = (raw, configDir, rcModes, inlineFileRefs = true) =
 };
 //#endregion
 //#region src/workflows/unified.yaml
-var unified_default = "# The unified gtd workflow template — gtd's BUILT-IN DEFAULT workflow. When no\n# `workflow:` key is configured anywhere in the cwd→home config chain,\n# src/Config.ts runs this template directly, so a state command works with no\n# configuration at all. `gtd init` seeds only the settings a project tunes\n# (vars.testCommand + a modes: formatting suggestion), never the workflow\n# itself. See STATES.md §10 for the full walkthrough.\n#\n# Compiled through the exact same `compileWorkflowConfig` a user's `.gtdrc`\n# `workflow:` key goes through — no privileged code path\n# (src/workflows/templates.ts does the compiling). Every content string below\n# is INLINE (no `./`-relative file references): the bundled template ships\n# inside a single-file build (`dist/gtd.bundle.mjs`), so its content must\n# already be resolved at BUILD time. `renderInitConfig` (src/workflows/\n# templates.ts) can MATERIALIZE this inline source into a `.gtdrc.json`\n# `workflow:` value — the way to eject/customize the machine, and the hermetic\n# test fixture — but `gtd init` itself no longer writes the workflow.\n#\n# ── Authored with SUB-MACHINES (Option C) ──────────────────────────────────\n#\n# This template is written with `submachines:`/`use:` (see src/Submachines.ts\n# and STATES.md). Expansion is a compile-time raw → raw pre-pass with NO runtime\n# footprint: the pure engine only ever sees the flat concrete states, and each\n# invocation's `as:` rename map reproduces the exact state names the flat\n# template used — so the compiled `WorkflowDefinition` is BYTE-IDENTICAL to the\n# frozen `unified.flat.yaml` (pinned by src/workflows/golden.test.ts). The\n# sub-machines split into DEDUP (a gate/loop authored once, invoked several\n# times) and ENCAPSULATION (a single-instance cluster grouped for reading):\n#\n#   * makeGreen (dedup ×2)    — the check/fix/escalate health triple shared by\n#                               the simple build tail and the advanced package\n#                               loop (identical script/prompt/message).\n#   * assertGreen (dedup ×3)  — the green-baseline gate (check + human blocked)\n#                               guarding the simple, advanced and review entries.\n#   * qaLoop (dedup ×2)       — the checkbox Q&A author/answer loop shared by the\n#                               product (adv-grilling) and technical (architecting)\n#                               phases of the advanced flow.\n#   * planLoop (encap)        — the simple flow's plan-iteration loop.\n#   * humanReview (encap)     — the shared review tail (reviewing → await-review →\n#                               review-deciding → feedback collect/build).\n#   * specGate (encap)        — the advanced flow's per-package agentic review.\n#   * packageLoop (encap)     — the advanced flow's per-package build queue.\n#\n# ── Four entry points behind a green-baseline gate, one shared tail ─────────\n#\n# EVERY entry first runs the test suite and only proceeds when it is GREEN — you\n# never start new work (or a review) on top of a red baseline. A red run halts\n# at a human `*-start-blocked` gate (fix the failures and re-run, exactly like\n# the `escalate` states), or is what `gtd fix` exists to repair.\n#\n# The initial `idle` state FORKS on which steering file the human creates,\n# routing into that flow's OWN start gate:\n#\n#   * .gtd/REQUIREMENTS.md -> `adv-start-check` -> the ADVANCED flow (product +\n#     technical Q&A, package decomposition, per-package parallel build +\n#     agentic spec-review).\n#   * anything else (e.g. .gtd/TODO.md) -> `start-check` -> the SIMPLE flow (one\n#     grilling Q&A loop, a monolithic build, no decomposition, no agentic\n#     review).\n#\n# A third entry, `gtd review <commitish>`, jumps to `review-start-check`\n# (reviewEntry) and, once green, to `reviewing` to review a foreign branch\n# through the same shared tail. A fourth, `gtd fix`, jumps to `fix-check`\n# (fixEntry): a red suite drops straight into the shared `fixing` loop and out\n# through the review + squash tail; a green suite is a no-op back to `idle`.\n#\n# All three converge at `reviewing` -> `await-review`. Ticking a REVIEW.md box\n# means only \"I reviewed this hunk\"; a COMMENT (not an unticked box) is what asks\n# for changes. Every human step at `await-review` goes to `review-deciding`,\n# which inspects the step and routes:\n#   * no comment (every box ticked, no REVIEW.md note, no code edit) -> the diff\n#     is a bare REVIEW.md deletion -> squashing -> done (the ONLY path to the\n#     finale — the whole cycle collapses into one commit).\n#   * any comment — a note in REVIEW.md OR a hand-edit to code — ->\n#     review-deciding writes REVIEW_FEEDBACK.md -> feedback-building implements\n#     it (completing any code the human hand-edited, never reverting their\n#     lines) -> checking -> reviewing regenerates an INCREMENTAL REVIEW.md\n#     (review-deciding carries `reviewBase: true`, so each feedback round's\n#     review + checkout window covers only `last-review..HEAD`, i.e. the agent's\n#     follow-through, not the human's own edit) -> await-review, looping until\n#     sign-off.\n# Two dead ends never commit: the sign-off gate (src/program.ts) REFUSES a step\n# that leaves a box unticked with no comment (finish reviewing first) and a\n# deleted REVIEW.md (restore it and tick the boxes).\n#\n# The shared health segment (`checking`/`fixing`/`escalate`) serves the simple\n# build and the feedback loop — its green edge targets\n# `reviewing`. The advanced per-package loop keeps its OWN health triple\n# (`adv-checking`/`adv-fixing`/`adv-escalate`) because its green edge targets\n# the per-package `spec-review` gate instead.\n#\n# Every `on` pattern's path below is written as the matching var's Eta\n# template (e.g. `\"A <%= it.vars.feedbackFile %>\"`), not a literal `.gtd/...`\n# path — repointing a var reroutes the machine along with every template that\n# reads/writes it (see docs/configuration.md).\n\nvars:\n  testCommand: npm test\n  todoFile: .gtd/TODO.md\n  requirementsFile: .gtd/REQUIREMENTS.md\n  architectureFile: .gtd/ARCHITECTURE.md\n  packagesDir: .gtd/packages\n  nextFile: .gtd/NEXT.md\n  reviewFile: .gtd/REVIEW.md\n  reviewRawFile: .gtd/REVIEW_RAW.md\n  reviewFeedbackFile: .gtd/REVIEW_FEEDBACK.md\n  feedbackFile: .gtd/FEEDBACK.md\n  specFeedbackFile: .gtd/SPEC_FEEDBACK.md\n  commitMsgFile: .gtd/COMMIT_MSG.md\n  # The two opaque model tiers every agent state draws from — repoint them here\n  # (or via a top-level `.gtdrc` `vars:` key / a `GTD_VAR_plannerModel` env var)\n  # to change every state on that tier at once. `plannerModel` (default \"smart\")\n  # is the heavier tier for the one-shot planning/architecting/review turns;\n  # `coderModel` (default \"base\") is the tier for the build/fix turns.\n  plannerModel: smart\n  coderModel: base\n\nsubmachines:\n  # ── DEDUP: the check/fix/escalate health triple ───────────────────────────\n  # Runs the suite; a red run records .gtd/FEEDBACK.md, a green run cleans it up.\n  # The `on` rules turn that into \"fix\" vs \"$onGreen\". The fix turn retries up to\n  # 3× before escalating to a human. Invoked for the shared build tail\n  # (checking/fixing/escalate -> reviewing) and the advanced package loop\n  # (adv-checking/adv-fixing/adv-escalate -> spec-review).\n  makeGreen:\n    params: [onGreen, escalateDescribe, checkLabel, fixLabel, escalateLabel]\n    states:\n      check:\n        actor: check\n        label: $checkLabel\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn — run the suite; a red run records .gtd/FEEDBACK.md, a\n          # green run cleans it up. The `on` rules decide what that means.\n          set +e\n          mkdir -p .gtd\n          # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n          # newline after an interpolation tag, so no tag may be the last token on\n          # a line — it would glue the next line's `else`/`fi` onto it and break\n          # the script. Below this point everything is plain bash.\n          feedback=\"<%~ it.vars.feedbackFile %>\"\n          <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n          code=$?\n          if [ \"$code\" -ne 0 ]; then\n            if [ -s .gtd/.check-output ]; then\n              mv .gtd/.check-output \"$feedback\"\n            else\n              rm -f .gtd/.check-output\n              printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n            fi\n          else\n            rm -f .gtd/.check-output\n            rm -f \"$feedback\"\n          fi\n        on:\n          \"A <%= it.vars.feedbackFile %>\": fix\n          \"M <%= it.vars.feedbackFile %>\": fix\n          \"D <%= it.vars.feedbackFile %>\": $onGreen\n          \"C\": $onGreen\n      fix:\n        actor: agent\n        label: $fixLabel\n        file: <%= it.vars.feedbackFile %>\n        model: <%= it.vars.coderModel %>\n        memory: fix\n        retry:\n          max: 3\n          otherwise: escalate\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.feedbackFile %>`, which this prompt tells you to address.\n\n          Read `<%= it.vars.feedbackFile %>` (the failing test output) and fix the\n          code so the suite passes. Keep the change focused — do not refactor\n          unrelated code. If the feedback is wrong, empty or delete\n          `<%= it.vars.feedbackFile %>` instead of \"fixing\" a non-issue.\n\n          Leave everything uncommitted and finish your turn — do not commit.\n        on:\n          \"* **\": check\n      escalate:\n        actor: human\n        label: $escalateLabel\n        file: <%= it.vars.feedbackFile %>\n        message: |\n          The agent could not get the check to pass after repeated attempts.\n          `<%= it.vars.feedbackFile %>` holds the last failing output.\n\n          What each change does next (then run `gtd step human`):\n          <% it.edges.forEach(function (e) { if (e.describe) { %>\n          <%~ \"- \" + e.describe + \"\\n\" %>\n          <% } }) %>\n        on:\n          \"* **\":\n            to: check\n            describe: $escalateDescribe\n\n  # ── DEDUP: the green-baseline gate ─────────────────────────────────────────\n  # Runs the suite before any work starts (its script is the health check's test\n  # wrapper — red -> .gtd/FEEDBACK.md, green -> remove it); the `on` rules turn\n  # that into \"halt at blocked\" vs \"proceed to $onGreen\". The script and blocked\n  # message/describe differ per entry, so they ride in as `with:` params.\n  assertGreen:\n    params: [onGreen, checkScript, blockedMessage, blockedDescribe, checkLabel, blockedLabel]\n    states:\n      check:\n        actor: check\n        label: $checkLabel\n        script: $checkScript\n        on:\n          \"A <%= it.vars.feedbackFile %>\": blocked\n          \"M <%= it.vars.feedbackFile %>\": blocked\n          \"D <%= it.vars.feedbackFile %>\": $onGreen\n          \"C\": $onGreen\n      blocked:\n        actor: human\n        label: $blockedLabel\n        file: <%= it.vars.feedbackFile %>\n        message: $blockedMessage\n        on:\n          \"* **\":\n            to: check\n            describe: $blockedDescribe\n\n  # ── DEDUP: the checkbox Q&A author/answer loop ─────────────────────────────\n  # The agent develops a plan file, surfacing `## Open Questions` as checkbox\n  # lists; the human answers (answerGate refuses an unanswered step) and either\n  # advances or sends it back. Invoked for the product (adv-grilling) and\n  # technical (architecting) phases of the advanced flow.\n  qaLoop:\n    params:\n      [\n        authorFile,\n        authorPrompt,\n        answerMessage,\n        onAdvance,\n        advanceDescribe,\n        editDescribe,\n        authorLabel,\n        answerLabel,\n      ]\n    states:\n      author:\n        actor: agent\n        label: $authorLabel\n        file: $authorFile\n        mode: qa\n        model: <%= it.vars.plannerModel %>\n        memory: plan\n        prompt: $authorPrompt\n        on:\n          \"* **\": answer\n      answer:\n        actor: human\n        label: $answerLabel\n        file: $authorFile\n        mode: qa\n        answerGate: true\n        message: $answerMessage\n        on:\n          \"C\":\n            to: $onAdvance\n            describe: $advanceDescribe\n          \"* **\":\n            to: author\n            describe: $editDescribe\n\n  # ── ENCAPSULATION: the simple flow's plan-iteration loop ───────────────────\n  planLoop:\n    params: [onAccept]\n    states:\n      planning:\n        actor: agent\n        label: Refining the plan\n        file: <%= it.vars.todoFile %>\n        model: <%= it.vars.plannerModel %>\n        memory: plan\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.todoFile %>`, which this prompt tells you to develop.\n\n          `<%= it.vars.todoFile %>` holds a short sketch of what to build — or, on a\n          return lap, the plan you already wrote plus the human's edits and comments\n          on it. Read it, explore the codebase, and develop it into a concrete\n          implementation plan: the files to change, the approach, and the steps to\n          take.\n\n          Do NOT ask questions and do NOT add an \"Open Questions\" section — decide\n          every open point yourself and write the plan as a set of concrete\n          decisions. Where a decision is genuinely a judgement call, state the\n          choice you made and, in one line, why. The human iterates on the plan by\n          editing it directly, not by answering questions.\n\n          On a return lap, `<%= it.vars.todoFile %>` carries the human's edits and\n          inline comments on your previous plan. Fold every edit and comment into a\n          revised, self-contained plan and remove the addressed comments — the plan\n          is always the current, clean proposal, never a diff.\n\n          Leave `<%= it.vars.todoFile %>` uncommitted and finish your turn — do\n          not commit, and do not run `gtd step agent` yourself; the harness does\n          that.\n        on:\n          \"* **\": plan-review\n\n      plan-review:\n        actor: human\n        label: Awaiting your review\n        file: <%= it.vars.todoFile %>\n        message: |\n          `<%= it.vars.todoFile %>` holds the proposed implementation plan. Read it,\n          then either accept it as-is or edit it — rewrite a step, add a\n          constraint, or drop an inline comment — to send it back for another round.\n\n          What each change does next (then run `gtd step human`):\n          <% it.edges.forEach(function (e) { if (e.describe) { %>\n          <%~ \"- \" + e.describe + \"\\n\" %>\n          <% } }) %>\n        on:\n          \"C\":\n            to: $onAccept\n            describe: >-\n              Change nothing to accept the plan as-is and move on to implementation\n              (**building**).\n          \"* **\":\n            to: planning\n            describe: >-\n              Edit the plan — rewrite a step, add a constraint, or leave an inline\n              comment — to send it back for another round (**planning**).\n\n  # ── ENCAPSULATION: the shared review tail ──────────────────────────────────\n  humanReview:\n    params: [onSignoff]\n    states:\n      reviewing:\n        actor: agent\n        label: Reviewing\n        file: <%= it.vars.reviewFile %>\n        mode: review\n        model: <%= it.vars.plannerModel %>\n        memory: review\n        # `gtd review <commitish>` (a plain non-gtd branch, e.g. a colleague's PR)\n        # does NOT enter here directly — it enters `review-start-check` (reviewEntry)\n        # first, which runs the green-baseline gate and, once green, transitions\n        # here. The `Gtd-Review-Base:` trailer rides on that entry commit (the\n        # process's oldest), so everything below still reuses the same\n        # reviewing/await-review/feedback machinery over <commitish>..HEAD.\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.reviewFile %>`, which this prompt tells you to write.\n\n          Write `<%= it.vars.reviewFile %>` to help a human review the diff below in\n          this EXACT format:\n\n          - First non-blank line: `# Review: <%= it.currentCommit.slice(0, 7) %>`\n          - Somewhere in the document: `<!-- base: <%= it.startCommit %> -->`\n          - At least one `## <Chunk Title>` heading, grouping hunks semantically\n            (same feature/refactor/fix, even across files) — a short explanation of\n            what changed and why, then one `- [ ] ./path/to/file.ts#42 — note` file\n            pointer per hunk (checkboxes are for the human to tick, not you; every\n            path is `./`-relative, with an optional `#line` and an optional\n            ` — note`).\n\n          The diff below is `it.reviewDiff` — on a feedback round it covers ONLY the\n          changes made since the previous review, so you review just what changed,\n          not items already signed off. Leave `<%= it.vars.reviewFile %>`\n          uncommitted and finish your turn.\n          <% if (it.reviewDiff.trim()) { %>\n\n          ## Full diff under review\n\n          ```diff\n          <%~ it.reviewDiff %>\n          ```\n          <% } %>\n        on:\n          \"* **\": await-review\n\n      await-review:\n        actor: human\n        label: Awaiting your review\n        file: <%= it.vars.reviewFile %>\n        mode: review\n        # While the cycle rests here, gtd opens a \"review checkout window\": raw HEAD\n        # is rewound to the review base (the most-recent `review-deciding` commit —\n        # so a feedback round shows only `last-review..HEAD` — or the process start\n        # on the first review) with the working tree untouched, surfacing the diff\n        # as ordinary uncommitted changes. It closes on the next invocation.\n        #\n        # Every human step here routes to `review-deciding`, which decides sign-off\n        # vs. feedback from the step's CONTENT (a comment, not a tick). Ticking a box\n        # only records \"I reviewed this hunk\". Two dead ends can't be told apart by a\n        # file-pattern edge, so the sign-off gate (`enforceReviewSignoffGate` in\n        # src/program.ts) catches them BEFORE the step commits and refuses: an\n        # unfinished review (a box still `- [ ]` with no comment) and a deleted\n        # REVIEW.md.\n        reviewWindow: true\n        message: |\n          `<%= it.vars.reviewFile %>` holds the review record for the cycle — one\n          `- [ ]` checkbox per reviewable item, grouped into chunks. Tick a box\n          (`- [x]`) as you review each hunk; ticking only records that you've read\n          it, it is not sign-off.\n\n          When you've been through the whole diff, run `gtd step human`:\n\n          - Tick EVERY box and leave no comment — no note in\n            `<%= it.vars.reviewFile %>`, no code edit — to sign off and collapse the\n            cycle into one commit (**squashing**).\n          - Leave a comment to request changes — a note on a\n            `<%= it.vars.reviewFile %>` line, or a direct code edit — to send it to a\n            build + re-review round (**review-deciding** → **feedback-building**). A\n            code edit counts as your own fix: the agent completes the follow-through\n            it implies and never reverts your lines.\n\n          Stepping with a box still unticked and no comment is refused (finish\n          reviewing first); so is deleting `<%= it.vars.reviewFile %>`.\n        on:\n          \"* **\": review-deciding\n\n      review-deciding:\n        actor: check\n        label: Reviewing\n        file: <%= it.vars.reviewFile %>\n        mode: review\n        # Anchors the INCREMENTAL review base: its most-recent in-process commit is\n        # the point the previous review round ended, so the next `reviewing` turn\n        # (and the review checkout window) covers only the changes since then. On\n        # the first review there is no such commit, so the base is the cycle start.\n        reviewBase: true\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn — decide the human's review step from its CONTENT. HEAD is\n          # the human's `await-review -> review-deciding` commit; HEAD^ is the\n          # `reviewing` commit that wrote the agent's original REVIEW.md (the sign-off\n          # gate refuses an incomplete step, so review-deciding is entered exactly\n          # once per round — nothing commits in between). It is FEEDBACK when the\n          # human left a comment: a note in REVIEW.md (any edit beyond a \"[ ]\"->\"[x]\"\n          # tick) OR a hand-edit to any non-.gtd/ file this round. Otherwise it is a\n          # clean sign-off. This turn only CAPTURES the raw material into REVIEW_RAW.md\n          # (the `feedback-collecting` agent then turns it into instructions); it does\n          # NOT interpret it. Either way, remove REVIEW.md; what the resulting diff\n          # MEANS is decided by this state's own `on` rules — the A/M REVIEW_RAW.md\n          # rows are declared BEFORE the D REVIEW.md row so a feedback round (which\n          # also deletes REVIEW.md) is captured as feedback, not mistaken for a\n          # sign-off.\n          set +e\n          mkdir -p .gtd\n          # Hoist the paths once, at the TOP: Eta's autoTrim eats the newline after\n          # an interpolation tag, so no tag may be the last token on a line — it\n          # would glue the next line onto it and break the script. Below this point\n          # everything is plain bash.\n          reviewFile=\"<%~ it.vars.reviewFile %>\"\n          reviewRawFile=\"<%~ it.vars.reviewRawFile %>\"\n          # Which non-.gtd/ files did the human hand-edit this round?\n          codeFiles=$(git diff-tree --no-commit-id --name-only -r HEAD | grep -v '^\\.gtd/')\n          # A note is any REVIEW.md change beyond a checkbox flip: normalize every\n          # \"[ ]\"/\"[x]\" to a placeholder, then compare the agent's original (HEAD^)\n          # against the human's version (HEAD).\n          notes=\"\"\n          before=$(git show \"HEAD^:$reviewFile\" 2>/dev/null | sed -E 's/\\[[ xX]\\]/[_]/g')\n          after=$(git show \"HEAD:$reviewFile\" 2>/dev/null | sed -E 's/\\[[ xX]\\]/[_]/g')\n          if [ \"$before\" != \"$after\" ]; then notes=1; fi\n          if [ -n \"$codeFiles\" ] || [ -n \"$notes\" ]; then\n            {\n              echo \"Raw review material captured for classification. The human reviewed the diff and left the feedback below; a downstream agent turns it into an instruction list. This file is machine-captured input, not instructions.\"\n              if [ -n \"$notes\" ]; then\n                echo\n                echo \"## Notes the human added to REVIEW.md this round\"\n                echo\n                echo '```diff'\n                git show HEAD --format= -- \"$reviewFile\"\n                echo '```'\n              fi\n              if [ -n \"$codeFiles\" ]; then\n                echo\n                echo \"## Files the human hand-edited this round (already committed)\"\n                echo\n                echo '```diff'\n                git show HEAD --format= -- $codeFiles\n                echo '```'\n              fi\n            } > \"$reviewRawFile\"\n          fi\n          rm -f \"$reviewFile\"\n        on:\n          \"A <%= it.vars.reviewRawFile %>\": feedback-collecting\n          \"M <%= it.vars.reviewRawFile %>\": feedback-collecting\n          \"D <%= it.vars.reviewFile %>\": $onSignoff\n\n      feedback-collecting:\n        actor: agent\n        label: Collecting your feedback\n        file: <%= it.vars.reviewFeedbackFile %>\n        model: <%= it.vars.plannerModel %>\n        memory: review\n        prompt: |\n          You are an autonomous coding agent turning raw review feedback into an\n          explicit instruction list for a builder. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.reviewRawFile %>` (delete it when done) and\n          `<%= it.vars.reviewFeedbackFile %>` (write it). Do NOT change any other\n          file — you classify, you do not build.\n\n          The raw review material is: <%~ it.read(it.vars.reviewRawFile) %>\n\n          Classify every piece of it into `<%= it.vars.reviewFeedbackFile %>` as a\n          flat, numbered instruction list. Each item is `N. ./path/to/file.ts#line —\n          <imperative>`. Read the two RAW sections and apply these rules exactly:\n\n          1. NOTES the human added to REVIEW.md — each note is a MANDATORY change,\n             regardless of whether its checkbox is ticked (ticking means only \"I read\n             this hunk\", never sign-off). Anchor each to the `./path#line` its review\n             item names (or the chunk it sits under). Emit it as a concrete\n             imperative.\n          2. Every COMMENT the human ADDED this round (a `+` line that is a code\n             comment, in the hand-edited files) is a MANDATORY instruction the human\n             is giving the builder — even a plain-prose one, even if its box is\n             ticked. Emit an imperative to satisfy it and note that the comment line\n             itself must be REMOVED once addressed. If a comment turns out to describe\n             behaviour the code already has, the instruction is still \"confirm the\n             code matches and remove the comment line\".\n          3. Every NON-comment code line the human hand-edited is an already-committed\n             INTENT, NOT a do-then-delete item. Emit an instruction that names those\n             files and says: their lines are final — complete ONLY the follow-through\n             they imply (callers, tests, types, docs) and NEVER revert or rewrite\n             them. Do not turn their own edit into a task.\n\n          Split each hand-edited hunk per line: comment `+` lines follow rule 2, the\n          surrounding real code follows rule 3.\n\n          If — and only if — nothing in the raw material is actionable (e.g. the human\n          left only an approving remark), write `<%= it.vars.reviewFeedbackFile %>`\n          containing exactly one line: `NOTHING ACTIONABLE — <one-sentence reason>`.\n          Never invent work to satisfy a non-actionable remark.\n\n          Delete `<%= it.vars.reviewRawFile %>`, leave everything else uncommitted,\n          and finish your turn.\n        on:\n          \"A <%= it.vars.reviewFeedbackFile %>\": feedback-building\n          \"M <%= it.vars.reviewFeedbackFile %>\": feedback-building\n\n      feedback-building:\n        actor: agent\n        label: Addressing your feedback\n        file: <%= it.vars.reviewFeedbackFile %>\n        model: <%= it.vars.coderModel %>\n        memory: build\n        # Refuse a work-free turn: a step whose only pending change is deleting the\n        # instructions file (the original \"captured then discarded\" bug). The edge\n        # gate (`enforceFeedbackProgressGate` in src/program.ts) reads this flag and\n        # exempts a `NOTHING ACTIONABLE` sentinel (a non-actionable review round,\n        # which legitimately makes no code change). The pure engine carries it inert.\n        requireProgress: true\n        prompt: |\n          You are an autonomous coding agent addressing review feedback. `.gtd/`\n          holds this workflow's own state — never create, edit, or delete anything\n          under `.gtd/` except `<%= it.vars.reviewFeedbackFile %>`, which this\n          prompt tells you to address and then delete.\n\n          The instructions to implement are: <%~ it.read(it.vars.reviewFeedbackFile) %>\n\n          This is a flat instruction list, not a diff — execute every item exactly as\n          written, no Q&A, no re-planning, using TDD discipline:\n\n          - An item that says the human's own lines are final: build ON TOP of them to\n            complete only the follow-through it names (callers, tests, types, docs).\n            NEVER revert or rewrite the lines they already committed.\n          - An item derived from a human comment: satisfy it AND delete that comment\n            line from the code (the human's review comments are transient — none of\n            them survive this round).\n\n          If the file's sole content is a `NOTHING ACTIONABLE` line, make no code\n          change; just delete the instructions file.\n\n          Delete `<%= it.vars.reviewFeedbackFile %>` once every item is addressed.\n          Leave everything else uncommitted and finish your turn. Deleting the\n          instructions file WITHOUT making the changes it asks for is refused — the\n          feedback must actually be addressed.\n        on:\n          \"* **\": checking\n\n  # ── ENCAPSULATION: the advanced flow's per-package agentic review ──────────\n  specGate:\n    params: [onApproved]\n    states:\n      spec-review:\n        actor: agent\n        label: Reviewing the package\n        model: <%= it.vars.plannerModel %>\n        memory: review\n        # Bounded per cycle: once entered `max` times, a transition into it is\n        # redirected to `closing` (force-close the package and move on — the\n        # unresolved concern is deferred to the shared human review at the tail,\n        # which sees the whole diff anyway). NOTE the cap is PROCESS-scoped, so it\n        # pools across packages — the same documented limitation the shared\n        # `fixing` cap has (see STATES.md / docs/configuration.md).\n        retry:\n          max: 3\n          otherwise: $onApproved\n        prompt: |\n          You are an autonomous coding agent reviewing a freshly-built work package\n          against its own spec. `.gtd/` holds this workflow's own state — never\n          create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.specFeedbackFile %>`, which this prompt tells you to write\n          when — and only when — you find problems.\n\n          The package spec is: <%~ it.read(it.vars.nextFile) %>\n\n          Read that spec, then verify the package's implementation (the diff below)\n          against it: are all its tasks done, acceptance criteria met, and the code\n          sound and consistent with the codebase?\n\n          - If it fully satisfies the spec, write NOTHING and finish your turn — a\n            clean turn is your approval.\n          - If it does not, write `<%= it.vars.specFeedbackFile %>` listing the\n            concrete problems for a fix turn to address. Be specific.\n\n          Do not fix anything yourself and do not delete the package file — a later\n          step owns that. Finish your turn.\n          <% if (it.processDiff.trim()) { %>\n\n          ## Diff under review\n\n          ```diff\n          <%~ it.processDiff %>\n          ```\n          <% } %>\n        on:\n          \"A <%= it.vars.specFeedbackFile %>\": spec-fix\n          \"M <%= it.vars.specFeedbackFile %>\": spec-fix\n          \"D <%= it.vars.specFeedbackFile %>\": $onApproved\n          \"C\": $onApproved\n\n      spec-fix:\n        actor: agent\n        label: Fixing review feedback\n        file: <%= it.vars.specFeedbackFile %>\n        model: <%= it.vars.coderModel %>\n        memory: fix\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.specFeedbackFile %>`, which this prompt tells you to address\n          and then delete.\n\n          Read `<%= it.vars.specFeedbackFile %>` (the reviewer's concerns about the\n          current package) and the package spec (`<%= it.read(it.vars.nextFile) %>`),\n          then fix the code to resolve every concern. Keep the change focused.\n\n          Delete `<%= it.vars.specFeedbackFile %>` once every concern is addressed.\n          Leave everything else uncommitted and finish your turn.\n        on:\n          \"* **\": adv-checking\n\n  # ── ENCAPSULATION: the advanced flow's per-package build queue ─────────────\n  packageLoop:\n    params: [onDrained]\n    states:\n      picking:\n        actor: check\n        label: Picking the next package\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn (the package queue arbiter) — the loop driver executes\n          # this verbatim, then steps the check actor. Mechanics only: take the\n          # first package file (by name) under the packages dir into NEXT.md, or\n          # remove NEXT.md when the queue is empty. What NEXT.md's presence/absence\n          # MEANS (another package to build vs. the queue closing out to review) is\n          # decided by this state's own `on` rules at capture time — never here.\n          set +e\n          # Hoist the NEXT path once, at the TOP: Eta's autoTrim eats the newline\n          # after an interpolation tag, so no tag may be the last token on a line —\n          # it would glue the next line's `else`/`fi` onto it and break the script.\n          # Below this point everything is plain bash.\n          nextFile=\"<%~ it.vars.nextFile %>\"\n          next=$(ls <%~ it.vars.packagesDir %>/*.md 2>/dev/null | head -n 1)\n          if [ -n \"$next\" ]; then\n            printf '%s' \"$next\" > \"$nextFile\"\n          else\n            rm -f \"$nextFile\"\n          fi\n        on:\n          \"D <%= it.vars.nextFile %>\": $onDrained\n          \"* <%= it.vars.nextFile %>\": adv-building\n          \"C\": $onDrained\n\n      adv-building:\n        actor: agent\n        label: Building\n        model: <%= it.vars.coderModel %>\n        memory: build\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/`; in\n          particular do NOT delete the package file (the `spec-review` gate reads it\n          after you) and never touch `<%= it.vars.nextFile %>` — the `picking` state\n          owns it.\n\n          The package to implement is: <%~ it.read(it.vars.nextFile) %>\n\n          That file describes a set of INDEPENDENT tasks. Implement ALL of them in\n          this one turn, fanning the independent tasks out to parallel subagents\n          where your harness supports it — each task is self-contained and can be\n          built concurrently. Implement EXACTLY the tasks this package describes — no\n          more, no less; leave every other package file untouched. Use TDD\n          discipline within each task.\n\n          Leave the package file in place and everything else uncommitted, then\n          finish your turn.\n        on:\n          \"* **\": adv-checking\n\n      closing:\n        actor: check\n        label: Closing out the package\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn (the package closer) — remove the just-reviewed package\n          # file (its path is in NEXT.md) plus any leftover spec feedback, so\n          # `picking` selects the NEXT package (or falls through to review when the\n          # queue is empty). Reached on approval (spec-review clean) and on the\n          # spec-review retry cap (force-close, deferring the concern to the tail).\n          set +e\n          pkg=$(cat <%~ it.vars.nextFile %> 2>/dev/null)\n          [ -n \"$pkg\" ] && rm -f \"$pkg\"\n          rm -f <%~ it.vars.specFeedbackFile %> <%~ it.vars.nextFile %>\n        on:\n          \"* **\": picking\n\nuse:\n  # Shared health triple: checking/fixing/escalate -> reviewing.\n  - submachine: makeGreen\n    name: build\n    as: { check: checking, fix: fixing, escalate: escalate }\n    with:\n      onGreen: reviewing\n      escalateDescribe: >-\n        Fix the failure yourself — edit the code and/or `.gtd/FEEDBACK.md` —\n        to re-run the check (**checking**).\n      checkLabel: Running checks\n      fixLabel: Fixing the check\n      escalateLabel: Escalating to a human\n  # Advanced package health triple: adv-checking/adv-fixing/adv-escalate -> spec-review.\n  - submachine: makeGreen\n    name: pkg\n    as: { check: adv-checking, fix: adv-fixing, escalate: adv-escalate }\n    with:\n      onGreen: spec-review\n      escalateDescribe: >-\n        Fix the failure yourself — edit the code and/or\n        `.gtd/FEEDBACK.md` — to re-run the check (**adv-checking**).\n      checkLabel: Running checks\n      fixLabel: Fixing the check\n      escalateLabel: Escalating to a human\n\n  # Green-baseline gate: simple flow entry.\n  - submachine: assertGreen\n    name: start\n    as: { check: start-check, blocked: start-blocked }\n    with:\n      onGreen: planning\n      checkScript: |\n        #!/usr/bin/env bash\n        # gtd check turn — run the suite before starting new work; a red run\n        # records .gtd/FEEDBACK.md (halt), a green run cleans it up (proceed).\n        set +e\n        mkdir -p .gtd\n        # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n        # newline after an interpolation tag, so no tag may be the last token on\n        # a line — it would glue the next line's `else`/`fi` onto it and break\n        # the script. Below this point everything is plain bash.\n        feedback=\"<%~ it.vars.feedbackFile %>\"\n        <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n        code=$?\n        if [ \"$code\" -ne 0 ]; then\n          if [ -s .gtd/.check-output ]; then\n            mv .gtd/.check-output \"$feedback\"\n          else\n            rm -f .gtd/.check-output\n            printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n          fi\n        else\n          rm -f .gtd/.check-output\n          rm -f \"$feedback\"\n        fi\n      blockedMessage: |\n        The test baseline is red — gtd will not start new work on a broken suite.\n        `<%= it.vars.feedbackFile %>` holds the failing output.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      blockedDescribe: >-\n        Fix the failing tests — edit the code and/or `.gtd/FEEDBACK.md` — to\n        re-run the baseline check (**start-check**). To repair the baseline as\n        its own separate reviewed commit instead, abandon this start and run\n        `gtd fix` from a clean `idle`.\n      checkLabel: Checking the baseline\n      blockedLabel: Baseline is red\n\n  # Green-baseline gate: advanced flow entry.\n  - submachine: assertGreen\n    name: adv-start\n    as: { check: adv-start-check, blocked: adv-start-blocked }\n    with:\n      onGreen: adv-grilling\n      checkScript: |\n        #!/usr/bin/env bash\n        # gtd check turn — run the suite before starting the advanced flow; a red\n        # run records .gtd/FEEDBACK.md (halt), a green run cleans it up (proceed).\n        set +e\n        mkdir -p .gtd\n        # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n        # newline after an interpolation tag, so no tag may be the last token on\n        # a line — it would glue the next line's `else`/`fi` onto it and break\n        # the script. Below this point everything is plain bash.\n        feedback=\"<%~ it.vars.feedbackFile %>\"\n        <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n        code=$?\n        if [ \"$code\" -ne 0 ]; then\n          if [ -s .gtd/.check-output ]; then\n            mv .gtd/.check-output \"$feedback\"\n          else\n            rm -f .gtd/.check-output\n            printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n          fi\n        else\n          rm -f .gtd/.check-output\n          rm -f \"$feedback\"\n        fi\n      blockedMessage: |\n        The test baseline is red — gtd will not start new work on a broken suite.\n        `<%= it.vars.feedbackFile %>` holds the failing output.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      blockedDescribe: >-\n        Fix the failing tests — edit the code and/or `.gtd/FEEDBACK.md` — to\n        re-run the baseline check (**adv-start-check**). To repair the baseline\n        as its own separate reviewed commit instead, abandon this start and run\n        `gtd fix` from a clean `idle`.\n      checkLabel: Checking the baseline\n      blockedLabel: Baseline is red\n\n  # Green-baseline gate: `gtd review` entry (reviewEntry -> reviewing).\n  - submachine: assertGreen\n    name: review-start\n    as: { check: review-start-check, blocked: review-start-blocked }\n    set:\n      check: { reviewEntry: true }\n    with:\n      onGreen: reviewing\n      checkScript: |\n        #!/usr/bin/env bash\n        # gtd check turn — run the suite before reviewing; a red run records\n        # .gtd/FEEDBACK.md (halt), a green run cleans it up (proceed to reviewing).\n        set +e\n        mkdir -p .gtd\n        # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n        # newline after an interpolation tag, so no tag may be the last token on\n        # a line — it would glue the next line's `else`/`fi` onto it and break\n        # the script. Below this point everything is plain bash.\n        feedback=\"<%~ it.vars.feedbackFile %>\"\n        <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n        code=$?\n        if [ \"$code\" -ne 0 ]; then\n          if [ -s .gtd/.check-output ]; then\n            mv .gtd/.check-output \"$feedback\"\n          else\n            rm -f .gtd/.check-output\n            printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n          fi\n        else\n          rm -f .gtd/.check-output\n          rm -f \"$feedback\"\n        fi\n      blockedMessage: |\n        The test baseline is red — gtd will not start a review on a broken suite.\n        `<%= it.vars.feedbackFile %>` holds the failing output.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      blockedDescribe: >-\n        Fix the failing tests — edit the code and/or `.gtd/FEEDBACK.md` — to\n        re-run the baseline check (**review-start-check**).\n      checkLabel: Checking the baseline\n      blockedLabel: Baseline is red\n\n  # Advanced flow: product Q&A loop (adv-grilling/adv-grilling-answer -> architecting).\n  - submachine: qaLoop\n    name: grill\n    as: { author: adv-grilling, answer: adv-grilling-answer }\n    with:\n      authorFile: <%= it.vars.requirementsFile %>\n      onAdvance: architecting\n      authorPrompt: |\n        You are an autonomous coding agent. `.gtd/` holds this workflow's own\n        state — never create, edit, or delete anything under `.gtd/` except\n        `<%= it.vars.requirementsFile %>`, which this prompt tells you to develop.\n\n        `<%= it.vars.requirementsFile %>` holds the product requirements (or the\n        plan you already wrote plus the human's answers on a return lap). Read it,\n        explore the codebase, and develop it into a concrete, PRODUCT-level plan\n        in this one turn.\n\n        Scope: product and user-facing decisions only. Leave implementation\n        details (file/module structure, data models, tech-stack choices) for the\n        next phase (`<%= it.vars.architectureFile %>`).\n\n        For every open point you cannot settle yourself, add a question under a\n        `## Open Questions` heading near the top: one `### <question>` sub-heading\n        each, followed by a checkbox list of candidate answers — aim for TWO\n        concrete options plus a final free-text slot, ALL left UNTICKED:\n\n            ### <the question>\n\n            - [ ] <first option — a concrete answer, a few words of rationale>\n            - [ ] <second option>\n            - [ ] _your answer_\n\n        Never tick a box yourself — the human ticks exactly one per question. The\n        last `- [ ] _your answer_` line is the human's own-answer slot; write it\n        with the literal `_your answer_` placeholder and leave it for them.\n\n        On a RETURN lap the human has ticked one option per question (or replaced\n        `_your answer_` with their own text). For each answered question: fold the\n        chosen answer into the plan prose, then MOVE the question down into a\n        `## Answered Questions` section as `### <question>` followed by the\n        resolved answer in plain PROSE (drop the checkboxes). Never re-raise a\n        question the human deleted; if the human deleted the whole\n        `## Open Questions` section, treat the plan as accepted and surface no new\n        questions unless something is genuinely still blocking. Omit\n        `## Open Questions` entirely once there are none left.\n\n        Leave `<%= it.vars.requirementsFile %>` uncommitted and finish your turn.\n      answerMessage: |\n        `<%= it.vars.requirementsFile %>` holds the product plan under\n        development. Each question under `## Open Questions` offers a few options\n        plus a `- [ ] _your answer_` slot. Answer EVERY question by ticking\n        exactly one box (`- [x]`); for your own answer, replace `_your answer_`\n        with your text and tick that line. Stepping is refused while any question\n        is unanswered.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      advanceDescribe: >-\n        With no open questions left (the agent surfaced none, or you deleted\n        the section and it finalized the plan), change nothing to accept the\n        product plan and move on to technical architecting (**architecting**).\n      editDescribe: >-\n        Tick exactly one option per open question (replace `_your answer_` for\n        your own) to send the plan back for the agent to fold your answers in\n        (**adv-grilling**). Delete a question to skip it, or delete the whole\n        `## Open Questions` section to accept the plan as-is.\n      authorLabel: Refining the product plan\n      answerLabel: Awaiting your answers\n\n  # Advanced flow: technical Q&A loop (architecting/architecting-answer -> decompose).\n  - submachine: qaLoop\n    name: arch\n    as: { author: architecting, answer: architecting-answer }\n    with:\n      authorFile: <%= it.vars.architectureFile %>\n      onAdvance: decompose\n      authorPrompt: |\n        You are an autonomous coding agent. `.gtd/` holds this workflow's own\n        state — never create, edit, or delete anything under `.gtd/` except\n        `<%= it.vars.architectureFile %>` and `<%= it.vars.requirementsFile %>`,\n        which this prompt tells you to write and delete respectively.\n\n        Read `<%= it.vars.requirementsFile %>` (the converged product plan) and\n        develop `<%= it.vars.architectureFile %>` from it in this one turn:\n        file/module structure, data models, library/tech-stack choices,\n        error-handling strategy — the *how*, building on the *what* already\n        settled. Do not re-open product-level decisions; treat them as settled.\n\n        For every open point you cannot settle yourself, add a question under a\n        `## Open Questions` heading exactly as the grilling phase does: one\n        `### <question>` each, followed by a checkbox list — TWO concrete options\n        plus a final free-text slot, ALL left UNTICKED:\n\n            ### <the question>\n\n            - [ ] <first option>\n            - [ ] <second option>\n            - [ ] _your answer_\n\n        Never tick a box yourself. On a RETURN lap the human has ticked one option\n        per question (or replaced `_your answer_` with their own text): fold each\n        chosen answer into the plan prose, then MOVE the question into a\n        `## Answered Questions` section as `### <question>` + the resolved answer\n        in plain PROSE (drop the checkboxes). Never re-raise a deleted question;\n        treat a deleted `## Open Questions` section as acceptance. Omit\n        `## Open Questions` once there are none left.\n\n        Once `<%= it.vars.architectureFile %>` is written, delete\n        `<%= it.vars.requirementsFile %>` — its content has been folded in and it\n        must not linger. Leave everything uncommitted and finish your turn.\n      answerMessage: |\n        `<%= it.vars.architectureFile %>` holds the technical plan under\n        development. Each question under `## Open Questions` offers a few options\n        plus a `- [ ] _your answer_` slot. Answer EVERY question by ticking\n        exactly one box (`- [x]`); for your own answer, replace `_your answer_`\n        with your text and tick that line. Stepping is refused while any question\n        is unanswered.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      advanceDescribe: >-\n        With no open questions left (the agent surfaced none, or you deleted\n        the section and it finalized the plan), change nothing to accept the\n        technical plan and move on to decomposing the work into packages\n        (**decompose**).\n      editDescribe: >-\n        Tick exactly one option per open question (replace `_your answer_` for\n        your own) to send the plan back for the agent to fold your answers in\n        (**architecting**). Delete a question to skip it, or delete the whole\n        `## Open Questions` section to accept the plan as-is.\n      authorLabel: Refining the technical plan\n      answerLabel: Awaiting your answers\n\n  # Simple flow plan-iteration loop (planning/plan-review -> building).\n  - submachine: planLoop\n    with:\n      onAccept: building\n\n  # Shared review tail (reviewing -> await-review -> ... -> squashing).\n  - submachine: humanReview\n    with:\n      onSignoff: squashing\n\n  # Advanced per-package agentic review (spec-review/spec-fix -> closing).\n  - submachine: specGate\n    with:\n      onApproved: closing\n\n  # Advanced per-package build queue (picking/adv-building/closing -> reviewing).\n  - submachine: packageLoop\n    with:\n      onDrained: reviewing\n\nstates:\n  # ── Shared entry ────────────────────────────────────────────────────────\n  idle:\n    actor: human\n    initial: true\n    label: Idle\n    # Every human-gate `message:` below closes with the same rendered \"What\n    # each change does next\" list, built from `it.edges` (the resting state's\n    # own `on` edges — each edge's `describe:` is the single source of truth).\n    message: |\n      No active gtd cycle.\n\n      To start one, create a steering file describing what you want built:\n\n      - `.gtd/TODO.md` — a short sketch — starts the SIMPLE flow (plan, build,\n        review).\n      - `.gtd/REQUIREMENTS.md` — the product requirements — starts the ADVANCED\n        flow (product + technical Q&A, package decomposition, parallel build,\n        per-package agentic review).\n\n      What each change does next (then run `gtd step human`):\n      <% it.edges.forEach(function (e) { if (e.describe) { %>\n      <%~ \"- \" + e.describe + \"\\n\" %>\n      <% } }) %>\n    on:\n      \"* <%= it.vars.requirementsFile %>\":\n        to: adv-start-check\n        describe: >-\n          Save the product requirements to `.gtd/REQUIREMENTS.md` to start the\n          ADVANCED flow — gtd first checks the test baseline is green, then the\n          agent develops them through product then technical Q&A, decomposes the\n          work into packages, and builds each in parallel (**adv-start-check**).\n      \"* **\":\n        to: start-check\n        describe: >-\n          Save a short sketch to `.gtd/TODO.md` (or any other change) to start\n          the SIMPLE flow — gtd first checks the test baseline is green, then the\n          agent reads it, explores the codebase, and develops it into a concrete\n          implementation plan (**start-check**).\n\n  # ── Simple flow: monolithic build (plan loop is the `planLoop` sub-machine) ─\n  building:\n    actor: agent\n    label: Building\n    file: <%= it.vars.todoFile %>\n    model: <%= it.vars.coderModel %>\n    memory: build\n    prompt: |\n      You are an autonomous coding agent. `.gtd/` holds this workflow's own\n      state — never create or edit anything under `.gtd/` except deleting\n      `<%= it.vars.todoFile %>`, which this prompt tells you to delete once its\n      work is complete.\n\n      The plan to implement is: <%~ it.read(it.vars.todoFile) %>\n\n      Read that plan and implement it in full. Use TDD discipline: one test,\n      then the implementation that passes it, then the next — never write all\n      the tests first.\n\n      Delete `<%= it.vars.todoFile %>` once the work it describes is complete\n      and verified. Leave everything else uncommitted and finish your turn.\n    on:\n      \"* **\": checking\n\n  # ── Advanced flow: decomposition (Q&A loops are the `qaLoop` sub-machine) ───\n  decompose:\n    actor: agent\n    label: Decomposing into packages\n    model: <%= it.vars.coderModel %>\n    memory: build\n    prompt: |\n      You are an autonomous coding agent. `.gtd/` holds this workflow's own\n      state — never create, edit, or delete anything under `.gtd/` except the\n      package files this prompt tells you to write (under\n      `<%= it.vars.packagesDir %>/`), and `<%= it.vars.architectureFile %>`,\n      which it tells you to delete.\n\n      Read `<%= it.vars.architectureFile %>` (the converged technical plan) and\n      decompose it into an ordered set of WORK PACKAGES, one file each under\n      `<%= it.vars.packagesDir %>/` (e.g. `<%= it.vars.packagesDir %>/01-name.md`,\n      `<%= it.vars.packagesDir %>/02-name.md`, ...). Order the packages so the\n      test suite can pass again after each one lands.\n\n      Each package file describes a set of INDEPENDENT tasks that can be\n      implemented concurrently without ordering constraints between them — a\n      single build turn will fan them out to parallel subagents. Give each task\n      a clear description, acceptance criteria as `- [ ]` checkboxes, and the\n      relevant paths. Also state the package's overall requirement so it can be\n      reviewed against its spec. Do NOT put ordering-dependent tasks in the same\n      package — split them across ordered packages instead.\n\n      No package file may reference any other `.gtd/` file. Once the package\n      files are written, delete `<%= it.vars.architectureFile %>`. Leave\n      everything uncommitted and finish your turn.\n    on:\n      \"* <%= it.vars.packagesDir %>/**\": picking\n\n  # ── Green-baseline gate: `gtd fix` entry ──────────────────────────────────\n  # `gtd fix` enters HERE (fixEntry): run the suite, then a red run drops\n  # straight into the shared `fixing` loop (-> checking -> review + squash\n  # tail); a green run is a no-op back to `idle` (nothing to fix).\n  fix-check:\n    actor: check\n    fixEntry: true\n    label: Checking the baseline\n    script: |\n      #!/usr/bin/env bash\n      # gtd check turn — run the suite to find what needs fixing; a red run\n      # records .gtd/FEEDBACK.md (-> fixing), a green run cleans it up (-> idle).\n      set +e\n      mkdir -p .gtd\n      # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n      # newline after an interpolation tag, so no tag may be the last token on\n      # a line — it would glue the next line's `else`/`fi` onto it and break\n      # the script. Below this point everything is plain bash.\n      feedback=\"<%~ it.vars.feedbackFile %>\"\n      <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n      code=$?\n      if [ \"$code\" -ne 0 ]; then\n        if [ -s .gtd/.check-output ]; then\n          mv .gtd/.check-output \"$feedback\"\n        else\n          rm -f .gtd/.check-output\n          printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n        fi\n      else\n        rm -f .gtd/.check-output\n        rm -f \"$feedback\"\n      fi\n    on:\n      \"A <%= it.vars.feedbackFile %>\": fixing\n      \"M <%= it.vars.feedbackFile %>\": fixing\n      \"D <%= it.vars.feedbackFile %>\": idle\n      \"C\": idle\n\n  # ── Shared tail: squash finale (review loop is the `humanReview` sub-machine) ─\n  squashing:\n    actor: agent\n    label: Squashing\n    file: <%= it.vars.commitMsgFile %>\n    model: <%= it.vars.coderModel %>\n    memory: build\n    prompt: |\n      You are an autonomous coding agent. The cycle is approved and done.\n      `.gtd/` holds this workflow's own state — never create, edit, or delete\n      anything under `.gtd/` except `<%= it.vars.commitMsgFile %>`, which this\n      prompt tells you to write.\n\n      Leave `<%= it.vars.commitMsgFile %>` uncommitted and finish your turn —\n      entering this file squashes everything this process retains into one\n      commit using its content as the message. The diff below is exactly what\n      that commit will contain: `it.retainedDiff`, the changes THIS process\n      made (its trace/retry boundary onward). For a normal build cycle that is\n      the whole feature; for a `gtd review` process it is only the fixes made\n      during the review — NOT the changeset that was under review.\n      <% if (it.retainedDiff.trim()) { %>\n\n      Write `<%= it.vars.commitMsgFile %>` with ONE conventional-commits message\n      for the changes below (subject line `type(scope): subject`, imperative\n      mood, <= 72 characters; a body explaining the why, trade-offs, and key\n      decisions). Plain text, no markdown wrapper.\n\n      ## Retained diff\n\n      ```diff\n      <%~ it.retainedDiff %>\n      ```\n      <% } else { %>\n\n      This process retained NO changes (a clean review sign-off with no fixes).\n      Write `<%= it.vars.commitMsgFile %>` with exactly this one line and\n      nothing else:\n\n      chore: human review\n      <% } %>\n    on:\n      \"A <%= it.vars.commitMsgFile %>\": done\n      \"M <%= it.vars.commitMsgFile %>\": done\n\n  done:\n    commit: |\n      <%~ it.read(it.vars.commitMsgFile) %>\n\n      Token cost: <%= it.processCost %>\n      <% it.processCostByModel.forEach(function(m){ %>\n      - <%= m.model %>: <%= m.cost %>\n      <% }) %>\n";
+var unified_default = "# The unified gtd workflow template — gtd's BUILT-IN DEFAULT workflow. When no\n# `workflow:` key is configured anywhere in the cwd→home config chain,\n# src/Config.ts runs this template directly, so a state command works with no\n# configuration at all. `gtd init` seeds only the settings a project tunes\n# (vars.testCommand + a modes: formatting suggestion), never the workflow\n# itself. See STATES.md §10 for the full walkthrough.\n#\n# Compiled through the exact same `compileWorkflowConfig` a user's `.gtdrc`\n# `workflow:` key goes through — no privileged code path\n# (src/workflows/templates.ts does the compiling). Every content string below\n# is INLINE (no `./`-relative file references): the bundled template ships\n# inside a single-file build (`dist/gtd.bundle.mjs`), so its content must\n# already be resolved at BUILD time. `renderInitConfig` (src/workflows/\n# templates.ts) can MATERIALIZE this inline source into a `.gtdrc.json`\n# `workflow:` value — the way to eject/customize the machine, and the hermetic\n# test fixture — but `gtd init` itself no longer writes the workflow.\n#\n# ── Authored with SUB-MACHINES (Option C) ──────────────────────────────────\n#\n# This template is written with `submachines:`/`use:` (see src/Submachines.ts\n# and STATES.md). Expansion is a compile-time raw → raw pre-pass with NO runtime\n# footprint: the pure engine only ever sees the flat concrete states, and each\n# invocation's `as:` rename map reproduces the exact state names the flat\n# template used — so the compiled `WorkflowDefinition` is BYTE-IDENTICAL to the\n# frozen `unified.flat.yaml` (pinned by src/workflows/golden.test.ts). The\n# sub-machines split into DEDUP (a gate/loop authored once, invoked several\n# times) and ENCAPSULATION (a single-instance cluster grouped for reading):\n#\n#   * makeGreen (dedup ×2)    — the check/fix/escalate health triple shared by\n#                               the simple build tail and the advanced package\n#                               loop (identical script/prompt/message).\n#   * assertGreen (dedup ×3)  — the green-baseline gate (check + human blocked)\n#                               guarding the simple, advanced and review entries.\n#   * qaLoop (dedup ×2)       — the checkbox Q&A author/answer loop shared by the\n#                               product (adv-grilling) and technical (architecting)\n#                               phases of the advanced flow.\n#   * planLoop (encap)        — the simple flow's plan-iteration loop.\n#   * humanReview (encap)     — the shared review tail (reviewing → await-review →\n#                               review-deciding → feedback collect/build).\n#   * specGate (encap)        — the advanced flow's per-package agentic review.\n#   * packageLoop (encap)     — the advanced flow's per-package build queue.\n#\n# ── Four entry points behind a green-baseline gate, one shared tail ─────────\n#\n# EVERY entry first runs the test suite and only proceeds when it is GREEN — you\n# never start new work (or a review) on top of a red baseline. A red run halts\n# at a human `*-start-blocked` gate (fix the failures and re-run, exactly like\n# the `escalate` states), or is what `gtd fix` exists to repair.\n#\n# The initial `idle` state FORKS on which steering file the human creates,\n# routing into that flow's OWN start gate:\n#\n#   * .gtd/REQUIREMENTS.md -> `adv-start-check` -> the ADVANCED flow (product +\n#     technical Q&A, package decomposition, per-package parallel build +\n#     agentic spec-review).\n#   * anything else (e.g. .gtd/TODO.md) -> `start-check` -> the SIMPLE flow (one\n#     grilling Q&A loop, a monolithic build, no decomposition, no agentic\n#     review).\n#\n# A third entry, `gtd review <commitish>`, jumps to `review-start-check`\n# (reviewEntry) and, once green, to `reviewing` to review a foreign branch\n# through the same shared tail. A fourth, `gtd fix`, jumps to `fix-check`\n# (fixEntry): a red suite drops straight into the shared `fixing` loop and out\n# through the review + squash tail; a green suite is a no-op back to `idle`.\n#\n# All three converge at `reviewing` -> `await-review`. Ticking a REVIEW.md box\n# means only \"I reviewed this hunk\"; a COMMENT (not an unticked box) is what asks\n# for changes. Every human step at `await-review` goes to `review-deciding`,\n# which inspects the step and routes:\n#   * no comment (every box ticked, no REVIEW.md note, no code edit) -> the diff\n#     is a bare REVIEW.md deletion -> squashing -> done (the ONLY path to the\n#     finale — the whole cycle collapses into one commit).\n#   * any comment — a note in REVIEW.md OR a hand-edit to code — ->\n#     review-deciding writes REVIEW_FEEDBACK.md -> feedback-building implements\n#     it (completing any code the human hand-edited, never reverting their\n#     lines) -> checking -> reviewing regenerates an INCREMENTAL REVIEW.md\n#     (review-deciding carries `reviewBase: true`, so each feedback round's\n#     review + checkout window covers only `last-review..HEAD`, i.e. the agent's\n#     follow-through, not the human's own edit) -> await-review, looping until\n#     sign-off.\n# Two dead ends never commit: the sign-off gate (src/program.ts) REFUSES a step\n# that leaves a box unticked with no comment (finish reviewing first) and a\n# deleted REVIEW.md (restore it and tick the boxes).\n#\n# The shared health segment (`checking`/`fixing`/`escalate`) serves the simple\n# build and the feedback loop — its green edge targets\n# `reviewing`. The advanced per-package loop keeps its OWN health triple\n# (`adv-checking`/`adv-fixing`/`adv-escalate`) because its green edge targets\n# the per-package `spec-review` gate instead.\n#\n# Every `on` pattern's path below is written as the matching var's Eta\n# template (e.g. `\"A <%= it.vars.feedbackFile %>\"`), not a literal `.gtd/...`\n# path — repointing a var reroutes the machine along with every template that\n# reads/writes it (see docs/configuration.md).\n\nvars:\n  testCommand: npm test\n  todoFile: .gtd/TODO.md\n  requirementsFile: .gtd/REQUIREMENTS.md\n  architectureFile: .gtd/ARCHITECTURE.md\n  packagesDir: .gtd/packages\n  nextFile: .gtd/NEXT.md\n  reviewFile: .gtd/REVIEW.md\n  reviewRawFile: .gtd/REVIEW_RAW.md\n  reviewFeedbackFile: .gtd/REVIEW_FEEDBACK.md\n  feedbackFile: .gtd/FEEDBACK.md\n  specFeedbackFile: .gtd/SPEC_FEEDBACK.md\n  commitMsgFile: .gtd/COMMIT_MSG.md\n  # The two opaque model tiers every agent state draws from — repoint them here\n  # (or via a top-level `.gtdrc` `vars:` key / a `GTD_VAR_plannerModel` env var)\n  # to change every state on that tier at once. `plannerModel` (default \"smart\")\n  # is the heavier tier for the one-shot planning/architecting/review turns;\n  # `coderModel` (default \"base\") is the tier for the build/fix turns.\n  plannerModel: smart\n  coderModel: base\n\nsubmachines:\n  # ── DEDUP: the check/fix/escalate health triple ───────────────────────────\n  # Runs the suite; a red run records .gtd/FEEDBACK.md, a green run cleans it up.\n  # The `on` rules turn that into \"fix\" vs \"$onGreen\". The fix turn retries up to\n  # 3× before escalating to a human. Invoked for the shared build tail\n  # (checking/fixing/escalate -> reviewing) and the advanced package loop\n  # (adv-checking/adv-fixing/adv-escalate -> spec-review).\n  makeGreen:\n    params: [onGreen, escalateDescribe, checkLabel, fixLabel, escalateLabel]\n    states:\n      check:\n        actor: check\n        label: $checkLabel\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn — run the suite; a red run records .gtd/FEEDBACK.md, a\n          # green run cleans it up. The `on` rules decide what that means.\n          set +e\n          mkdir -p .gtd\n          # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n          # newline after an interpolation tag, so no tag may be the last token on\n          # a line — it would glue the next line's `else`/`fi` onto it and break\n          # the script. Below this point everything is plain bash.\n          feedback=\"<%~ it.vars.feedbackFile %>\"\n          <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n          code=$?\n          if [ \"$code\" -ne 0 ]; then\n            if [ -s .gtd/.check-output ]; then\n              mv .gtd/.check-output \"$feedback\"\n            else\n              rm -f .gtd/.check-output\n              printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n            fi\n          else\n            rm -f .gtd/.check-output\n            rm -f \"$feedback\"\n          fi\n        on:\n          \"A <%= it.vars.feedbackFile %>\": fix\n          \"M <%= it.vars.feedbackFile %>\": fix\n          \"D <%= it.vars.feedbackFile %>\": $onGreen\n          \"C\": $onGreen\n      fix:\n        actor: agent\n        label: $fixLabel\n        file: <%= it.vars.feedbackFile %>\n        model: <%= it.vars.coderModel %>\n        memory: fix\n        retry:\n          max: 3\n          otherwise: escalate\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.feedbackFile %>`, which this prompt tells you to address.\n\n          Read `<%= it.vars.feedbackFile %>` (the failing test output) and fix the\n          code so the suite passes. Keep the change focused — do not refactor\n          unrelated code. If the feedback is wrong, empty or delete\n          `<%= it.vars.feedbackFile %>` instead of \"fixing\" a non-issue.\n\n          Leave everything uncommitted and finish your turn — do not commit.\n        on:\n          \"* **\": check\n      escalate:\n        actor: human\n        label: $escalateLabel\n        file: <%= it.vars.feedbackFile %>\n        message: |\n          The agent could not get the check to pass after repeated attempts.\n          `<%= it.vars.feedbackFile %>` holds the last failing output.\n\n          What each change does next (then run `gtd step human`):\n          <% it.edges.forEach(function (e) { if (e.describe) { %>\n          <%~ \"- \" + e.describe + \"\\n\" %>\n          <% } }) %>\n        on:\n          \"* **\":\n            to: check\n            describe: $escalateDescribe\n\n  # ── DEDUP: the green-baseline gate ─────────────────────────────────────────\n  # Runs the suite before any work starts (its script is the health check's test\n  # wrapper — red -> .gtd/FEEDBACK.md, green -> remove it); the `on` rules turn\n  # that into \"halt at blocked\" vs \"proceed to $onGreen\". The script and blocked\n  # message/describe differ per entry, so they ride in as `with:` params.\n  assertGreen:\n    params: [onGreen, checkScript, blockedMessage, blockedDescribe, checkLabel, blockedLabel]\n    states:\n      check:\n        actor: check\n        label: $checkLabel\n        script: $checkScript\n        on:\n          \"A <%= it.vars.feedbackFile %>\": blocked\n          \"M <%= it.vars.feedbackFile %>\": blocked\n          \"D <%= it.vars.feedbackFile %>\": $onGreen\n          \"C\": $onGreen\n      blocked:\n        actor: human\n        label: $blockedLabel\n        file: <%= it.vars.feedbackFile %>\n        message: $blockedMessage\n        on:\n          \"* **\":\n            to: check\n            describe: $blockedDescribe\n\n  # ── DEDUP: the checkbox Q&A author/answer loop ─────────────────────────────\n  # The agent develops a plan file, surfacing `## Open Questions` as checkbox\n  # lists; the human answers (answerGate refuses an unanswered step) and either\n  # advances or sends it back. Invoked for the product (adv-grilling) and\n  # technical (architecting) phases of the advanced flow.\n  qaLoop:\n    params:\n      [\n        authorFile,\n        authorPrompt,\n        answerMessage,\n        onAdvance,\n        advanceDescribe,\n        editDescribe,\n        authorLabel,\n        answerLabel,\n      ]\n    states:\n      author:\n        actor: agent\n        label: $authorLabel\n        file: $authorFile\n        mode: qa\n        model: <%= it.vars.plannerModel %>\n        memory: plan\n        prompt: $authorPrompt\n        on:\n          \"* **\": answer\n      answer:\n        actor: human\n        label: $answerLabel\n        file: $authorFile\n        mode: qa\n        answerGate: true\n        message: $answerMessage\n        on:\n          \"C\":\n            to: $onAdvance\n            describe: $advanceDescribe\n          \"* **\":\n            to: author\n            describe: $editDescribe\n\n  # ── ENCAPSULATION: the simple flow's plan-iteration loop ───────────────────\n  planLoop:\n    params: [onAccept]\n    states:\n      planning:\n        actor: agent\n        label: Refining the plan\n        file: <%= it.vars.todoFile %>\n        model: <%= it.vars.plannerModel %>\n        memory: plan\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.todoFile %>`, which this prompt tells you to develop.\n\n          `<%= it.vars.todoFile %>` holds a short sketch of what to build — or, on a\n          return lap, the plan you already wrote plus the human's edits and comments\n          on it. Read it, explore the codebase, and develop it into a concrete\n          implementation plan: the files to change, the approach, and the steps to\n          take.\n\n          Do NOT ask questions and do NOT add an \"Open Questions\" section — decide\n          every open point yourself and write the plan as a set of concrete\n          decisions. Where a decision is genuinely a judgement call, state the\n          choice you made and, in one line, why. The human iterates on the plan by\n          editing it directly, not by answering questions.\n\n          On a return lap, `<%= it.vars.todoFile %>` carries the human's edits and\n          inline comments on your previous plan. Fold every edit and comment into a\n          revised, self-contained plan and remove the addressed comments — the plan\n          is always the current, clean proposal, never a diff.\n\n          Leave `<%= it.vars.todoFile %>` uncommitted and finish your turn — do\n          not commit, and do not run `gtd step agent` yourself; the harness does\n          that.\n        on:\n          \"* **\": plan-review\n\n      plan-review:\n        actor: human\n        label: Awaiting your review\n        file: <%= it.vars.todoFile %>\n        message: |\n          `<%= it.vars.todoFile %>` holds the proposed implementation plan. Read it,\n          then either accept it as-is or edit it — rewrite a step, add a\n          constraint, or drop an inline comment — to send it back for another round.\n\n          What each change does next (then run `gtd step human`):\n          <% it.edges.forEach(function (e) { if (e.describe) { %>\n          <%~ \"- \" + e.describe + \"\\n\" %>\n          <% } }) %>\n        on:\n          \"C\":\n            to: $onAccept\n            describe: >-\n              Change nothing to accept the plan as-is and move on to implementation\n              (**building**).\n          \"* **\":\n            to: planning\n            describe: >-\n              Edit the plan — rewrite a step, add a constraint, or leave an inline\n              comment — to send it back for another round (**planning**).\n\n  # ── ENCAPSULATION: the shared review tail ──────────────────────────────────\n  humanReview:\n    params: [onSignoff]\n    states:\n      reviewing:\n        actor: agent\n        label: Reviewing\n        file: <%= it.vars.reviewFile %>\n        mode: review\n        model: <%= it.vars.plannerModel %>\n        memory: review\n        # `gtd review <commitish>` (a plain non-gtd branch, e.g. a colleague's PR)\n        # does NOT enter here directly — it enters `review-start-check` (reviewEntry)\n        # first, which runs the green-baseline gate and, once green, transitions\n        # here. The `Gtd-Review-Base:` trailer rides on that entry commit (the\n        # process's oldest), so everything below still reuses the same\n        # reviewing/await-review/feedback machinery over <commitish>..HEAD.\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.reviewFile %>`, which this prompt tells you to write.\n\n          Write `<%= it.vars.reviewFile %>` to help a human review the diff below in\n          this EXACT format:\n\n          - First non-blank line: `# Review: <%= it.currentCommit.slice(0, 7) %>`\n          - Somewhere in the document: `<!-- base: <%= it.startCommit %> -->`\n          - At least one `## <Chunk Title>` heading, grouping hunks semantically\n            (same feature/refactor/fix, even across files) — a short explanation of\n            what changed and why, then one `- [ ] ./path/to/file.ts#42 — note` file\n            pointer per hunk (checkboxes are for the human to tick, not you; every\n            path is `./`-relative, with an optional `#line` and an optional\n            ` — note`).\n\n          The diff below is `it.reviewDiff` — on a feedback round it covers ONLY the\n          changes made since the previous review, so you review just what changed,\n          not items already signed off. Leave `<%= it.vars.reviewFile %>`\n          uncommitted and finish your turn.\n          <% if (it.reviewDiff.trim()) { %>\n\n          ## Full diff under review\n\n          ```diff\n          <%~ it.reviewDiff %>\n          ```\n          <% } %>\n        on:\n          \"* **\": await-review\n\n      await-review:\n        actor: human\n        label: Awaiting your review\n        file: <%= it.vars.reviewFile %>\n        mode: review\n        # While the cycle rests here, gtd opens a \"review checkout window\": raw HEAD\n        # is rewound to the review base (the most-recent `review-deciding` commit —\n        # so a feedback round shows only `last-review..HEAD` — or the process start\n        # on the first review) with the working tree untouched, surfacing the diff\n        # as ordinary uncommitted changes. It closes on the next invocation.\n        #\n        # Every human step here routes to `review-deciding`, which decides sign-off\n        # vs. feedback from the step's CONTENT (a comment, not a tick). Ticking a box\n        # only records \"I reviewed this hunk\". Two dead ends can't be told apart by a\n        # file-pattern edge, so the sign-off gate (`enforceReviewSignoffGate` in\n        # src/program.ts) catches them BEFORE the step commits and refuses: an\n        # unfinished review (a box still `- [ ]` with no comment) and a deleted\n        # REVIEW.md.\n        reviewWindow: true\n        message: |\n          `<%= it.vars.reviewFile %>` holds the review record for the cycle — one\n          `- [ ]` checkbox per reviewable item, grouped into chunks. Tick a box\n          (`- [x]`) as you review each hunk; ticking only records that you've read\n          it, it is not sign-off.\n\n          When you've been through the whole diff, run `gtd step human`:\n\n          - Tick EVERY box and leave no comment — no note in\n            `<%= it.vars.reviewFile %>`, no code edit — to sign off and collapse the\n            cycle into one commit (**squashing**).\n          - Leave a comment to request changes — a note on a\n            `<%= it.vars.reviewFile %>` line, or a direct code edit — to send it to a\n            build + re-review round (**review-deciding** → **feedback-building**). A\n            code edit counts as your own fix: the agent completes the follow-through\n            it implies and never reverts your lines.\n\n          Stepping with a box still unticked and no comment is refused (finish\n          reviewing first); so is deleting `<%= it.vars.reviewFile %>`.\n        on:\n          \"* **\": review-deciding\n\n      review-deciding:\n        actor: check\n        label: Reviewing\n        file: <%= it.vars.reviewFile %>\n        mode: review\n        # Anchors the INCREMENTAL review base: its most-recent in-process commit is\n        # the point the previous review round ended, so the next `reviewing` turn\n        # (and the review checkout window) covers only the changes since then. On\n        # the first review there is no such commit, so the base is the cycle start.\n        reviewBase: true\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn — decide the human's review step from its CONTENT. HEAD is\n          # the human's `await-review -> review-deciding` commit; HEAD^ is the\n          # `reviewing` commit that wrote the agent's original REVIEW.md (the sign-off\n          # gate refuses an incomplete step, so review-deciding is entered exactly\n          # once per round — nothing commits in between). It is FEEDBACK when the\n          # human left a comment: a note in REVIEW.md (any edit beyond a \"[ ]\"->\"[x]\"\n          # tick) OR a hand-edit to any file this round other than `.gtd/` or the\n          # state's own reviewFile (which is var-configurable and may live outside\n          # `.gtd/`, e.g. a root-level REVIEW.md — issue #128). Otherwise it is a\n          # clean sign-off. This turn only CAPTURES the raw material into REVIEW_RAW.md\n          # (the `feedback-collecting` agent then turns it into instructions); it does\n          # NOT interpret it. Either way, remove REVIEW.md; what the resulting diff\n          # MEANS is decided by this state's own `on` rules — the A/M REVIEW_RAW.md\n          # rows are declared BEFORE the D REVIEW.md row so a feedback round (which\n          # also deletes REVIEW.md) is captured as feedback, not mistaken for a\n          # sign-off.\n          set +e\n          mkdir -p .gtd\n          # Hoist the paths once, at the TOP: Eta's autoTrim eats the newline after\n          # an interpolation tag, so no tag may be the last token on a line — it\n          # would glue the next line onto it and break the script. Below this point\n          # everything is plain bash.\n          reviewFile=\"<%~ it.vars.reviewFile %>\"\n          reviewRawFile=\"<%~ it.vars.reviewRawFile %>\"\n          # Which files did the human hand-edit this round, besides .gtd/ and the\n          # reviewFile itself (var-configurable — may live outside .gtd/, so it is\n          # excluded by exact path, not just the .gtd/ prefix)?\n          codeFiles=$(git diff-tree --no-commit-id --name-only -r HEAD \\\n            | grep -v '^\\.gtd/' | grep -vxF \"$reviewFile\")\n          # A note is any REVIEW.md change beyond a checkbox flip: normalize every\n          # \"[ ]\"/\"[x]\" to a placeholder, then compare the agent's original (HEAD^)\n          # against the human's version (HEAD).\n          notes=\"\"\n          before=$(git show \"HEAD^:$reviewFile\" 2>/dev/null | sed -E 's/\\[[ xX]\\]/[_]/g')\n          after=$(git show \"HEAD:$reviewFile\" 2>/dev/null | sed -E 's/\\[[ xX]\\]/[_]/g')\n          if [ \"$before\" != \"$after\" ]; then notes=1; fi\n          if [ -n \"$codeFiles\" ] || [ -n \"$notes\" ]; then\n            {\n              echo \"Raw review material captured for classification. The human reviewed the diff and left the feedback below; a downstream agent turns it into an instruction list. This file is machine-captured input, not instructions.\"\n              if [ -n \"$notes\" ]; then\n                echo\n                echo \"## Notes the human added to REVIEW.md this round\"\n                echo\n                echo '```diff'\n                git show HEAD --format= -- \"$reviewFile\"\n                echo '```'\n              fi\n              if [ -n \"$codeFiles\" ]; then\n                echo\n                echo \"## Files the human hand-edited this round (already committed)\"\n                echo\n                echo '```diff'\n                git show HEAD --format= -- $codeFiles\n                echo '```'\n              fi\n            } > \"$reviewRawFile\"\n          fi\n          rm -f \"$reviewFile\"\n        on:\n          \"A <%= it.vars.reviewRawFile %>\": feedback-collecting\n          \"M <%= it.vars.reviewRawFile %>\": feedback-collecting\n          \"D <%= it.vars.reviewFile %>\": $onSignoff\n\n      feedback-collecting:\n        actor: agent\n        label: Collecting your feedback\n        file: <%= it.vars.reviewFeedbackFile %>\n        model: <%= it.vars.plannerModel %>\n        memory: review\n        prompt: |\n          You are an autonomous coding agent turning raw review feedback into an\n          explicit instruction list for a builder. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.reviewRawFile %>` (delete it when done) and\n          `<%= it.vars.reviewFeedbackFile %>` (write it). Do NOT change any other\n          file — you classify, you do not build.\n\n          The raw review material is: <%~ it.read(it.vars.reviewRawFile) %>\n\n          Classify every piece of it into `<%= it.vars.reviewFeedbackFile %>` as a\n          flat, numbered instruction list. Each item is `N. ./path/to/file.ts#line —\n          <imperative>`. Read the two RAW sections and apply these rules exactly:\n\n          1. NOTES the human added to REVIEW.md — each note is a MANDATORY change,\n             regardless of whether its checkbox is ticked (ticking means only \"I read\n             this hunk\", never sign-off). Anchor each to the `./path#line` its review\n             item names (or the chunk it sits under). Emit it as a concrete\n             imperative.\n          2. Every COMMENT the human ADDED this round (a `+` line that is a code\n             comment, in the hand-edited files) is a MANDATORY instruction the human\n             is giving the builder — even a plain-prose one, even if its box is\n             ticked. Emit an imperative to satisfy it and note that the comment line\n             itself must be REMOVED once addressed. If a comment turns out to describe\n             behaviour the code already has, the instruction is still \"confirm the\n             code matches and remove the comment line\".\n          3. Every NON-comment code line the human hand-edited is an already-committed\n             INTENT, NOT a do-then-delete item. Emit an instruction that names those\n             files and says: their lines are final — complete ONLY the follow-through\n             they imply (callers, tests, types, docs) and NEVER revert or rewrite\n             them. Do not turn their own edit into a task.\n\n          Split each hand-edited hunk per line: comment `+` lines follow rule 2, the\n          surrounding real code follows rule 3.\n\n          If — and only if — nothing in the raw material is actionable (e.g. the human\n          left only an approving remark), write `<%= it.vars.reviewFeedbackFile %>`\n          containing exactly one line: `NOTHING ACTIONABLE — <one-sentence reason>`.\n          Never invent work to satisfy a non-actionable remark.\n\n          Delete `<%= it.vars.reviewRawFile %>`, leave everything else uncommitted,\n          and finish your turn.\n        on:\n          \"A <%= it.vars.reviewFeedbackFile %>\": feedback-building\n          \"M <%= it.vars.reviewFeedbackFile %>\": feedback-building\n\n      feedback-building:\n        actor: agent\n        label: Addressing your feedback\n        file: <%= it.vars.reviewFeedbackFile %>\n        model: <%= it.vars.coderModel %>\n        memory: build\n        # Refuse a work-free turn: a step whose only pending change is deleting the\n        # instructions file (the original \"captured then discarded\" bug). The edge\n        # gate (`enforceFeedbackProgressGate` in src/program.ts) reads this flag and\n        # exempts a `NOTHING ACTIONABLE` sentinel (a non-actionable review round,\n        # which legitimately makes no code change). The pure engine carries it inert.\n        requireProgress: true\n        prompt: |\n          You are an autonomous coding agent addressing review feedback. `.gtd/`\n          holds this workflow's own state — never create, edit, or delete anything\n          under `.gtd/` except `<%= it.vars.reviewFeedbackFile %>`, which this\n          prompt tells you to address and then delete.\n\n          The instructions to implement are: <%~ it.read(it.vars.reviewFeedbackFile) %>\n\n          This is a flat instruction list, not a diff — execute every item exactly as\n          written, no Q&A, no re-planning, using TDD discipline:\n\n          - An item that says the human's own lines are final: build ON TOP of them to\n            complete only the follow-through it names (callers, tests, types, docs).\n            NEVER revert or rewrite the lines they already committed.\n          - An item derived from a human comment: satisfy it AND delete that comment\n            line from the code (the human's review comments are transient — none of\n            them survive this round).\n\n          If the file's sole content is a `NOTHING ACTIONABLE` line, make no code\n          change; just delete the instructions file.\n\n          Delete `<%= it.vars.reviewFeedbackFile %>` once every item is addressed.\n          Leave everything else uncommitted and finish your turn. Deleting the\n          instructions file WITHOUT making the changes it asks for is refused — the\n          feedback must actually be addressed.\n        on:\n          \"* **\": checking\n\n  # ── ENCAPSULATION: the advanced flow's per-package agentic review ──────────\n  specGate:\n    params: [onApproved]\n    states:\n      spec-review:\n        actor: agent\n        label: Reviewing the package\n        model: <%= it.vars.plannerModel %>\n        memory: review\n        # Bounded per cycle: once entered `max` times, a transition into it is\n        # redirected to `closing` (force-close the package and move on — the\n        # unresolved concern is deferred to the shared human review at the tail,\n        # which sees the whole diff anyway). NOTE the cap is PROCESS-scoped, so it\n        # pools across packages — the same documented limitation the shared\n        # `fixing` cap has (see STATES.md / docs/configuration.md).\n        retry:\n          max: 3\n          otherwise: $onApproved\n        prompt: |\n          You are an autonomous coding agent reviewing a freshly-built work package\n          against its own spec. `.gtd/` holds this workflow's own state — never\n          create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.specFeedbackFile %>`, which this prompt tells you to write\n          when — and only when — you find problems.\n\n          The package spec is: <%~ it.read(it.vars.nextFile) %>\n\n          Read that spec, then verify the package's implementation (the diff below)\n          against it: are all its tasks done, acceptance criteria met, and the code\n          sound and consistent with the codebase?\n\n          - If it fully satisfies the spec, write NOTHING and finish your turn — a\n            clean turn is your approval.\n          - If it does not, write `<%= it.vars.specFeedbackFile %>` listing the\n            concrete problems for a fix turn to address. Be specific.\n\n          Do not fix anything yourself and do not delete the package file — a later\n          step owns that. Finish your turn.\n          <% if (it.processDiff.trim()) { %>\n\n          ## Diff under review\n\n          ```diff\n          <%~ it.processDiff %>\n          ```\n          <% } %>\n        on:\n          \"A <%= it.vars.specFeedbackFile %>\": spec-fix\n          \"M <%= it.vars.specFeedbackFile %>\": spec-fix\n          \"D <%= it.vars.specFeedbackFile %>\": $onApproved\n          \"C\": $onApproved\n\n      spec-fix:\n        actor: agent\n        label: Fixing review feedback\n        file: <%= it.vars.specFeedbackFile %>\n        model: <%= it.vars.coderModel %>\n        memory: fix\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/` except\n          `<%= it.vars.specFeedbackFile %>`, which this prompt tells you to address\n          and then delete.\n\n          Read `<%= it.vars.specFeedbackFile %>` (the reviewer's concerns about the\n          current package) and the package spec (`<%= it.read(it.vars.nextFile) %>`),\n          then fix the code to resolve every concern. Keep the change focused.\n\n          Delete `<%= it.vars.specFeedbackFile %>` once every concern is addressed.\n          Leave everything else uncommitted and finish your turn.\n        on:\n          \"* **\": adv-checking\n\n  # ── ENCAPSULATION: the advanced flow's per-package build queue ─────────────\n  packageLoop:\n    params: [onDrained]\n    states:\n      picking:\n        actor: check\n        label: Picking the next package\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn (the package queue arbiter) — the loop driver executes\n          # this verbatim, then steps the check actor. Mechanics only: take the\n          # first package file (by name) under the packages dir into NEXT.md, or\n          # remove NEXT.md when the queue is empty. What NEXT.md's presence/absence\n          # MEANS (another package to build vs. the queue closing out to review) is\n          # decided by this state's own `on` rules at capture time — never here.\n          set +e\n          # Hoist the NEXT path once, at the TOP: Eta's autoTrim eats the newline\n          # after an interpolation tag, so no tag may be the last token on a line —\n          # it would glue the next line's `else`/`fi` onto it and break the script.\n          # Below this point everything is plain bash.\n          nextFile=\"<%~ it.vars.nextFile %>\"\n          next=$(ls <%~ it.vars.packagesDir %>/*.md 2>/dev/null | head -n 1)\n          if [ -n \"$next\" ]; then\n            printf '%s' \"$next\" > \"$nextFile\"\n          else\n            rm -f \"$nextFile\"\n          fi\n        on:\n          \"D <%= it.vars.nextFile %>\": $onDrained\n          \"* <%= it.vars.nextFile %>\": adv-building\n          \"C\": $onDrained\n\n      adv-building:\n        actor: agent\n        label: Building\n        model: <%= it.vars.coderModel %>\n        memory: build\n        prompt: |\n          You are an autonomous coding agent. `.gtd/` holds this workflow's own\n          state — never create, edit, or delete anything under `.gtd/`; in\n          particular do NOT delete the package file (the `spec-review` gate reads it\n          after you) and never touch `<%= it.vars.nextFile %>` — the `picking` state\n          owns it.\n\n          The package to implement is: <%~ it.read(it.vars.nextFile) %>\n\n          That file describes a set of INDEPENDENT tasks. Implement ALL of them in\n          this one turn, fanning the independent tasks out to parallel subagents\n          where your harness supports it — each task is self-contained and can be\n          built concurrently. Implement EXACTLY the tasks this package describes — no\n          more, no less; leave every other package file untouched. Use TDD\n          discipline within each task.\n\n          Leave the package file in place and everything else uncommitted, then\n          finish your turn.\n        on:\n          \"* **\": adv-checking\n\n      closing:\n        actor: check\n        label: Closing out the package\n        script: |\n          #!/usr/bin/env bash\n          # gtd check turn (the package closer) — remove the just-reviewed package\n          # file (its path is in NEXT.md) plus any leftover spec feedback, so\n          # `picking` selects the NEXT package (or falls through to review when the\n          # queue is empty). Reached on approval (spec-review clean) and on the\n          # spec-review retry cap (force-close, deferring the concern to the tail).\n          set +e\n          pkg=$(cat <%~ it.vars.nextFile %> 2>/dev/null)\n          [ -n \"$pkg\" ] && rm -f \"$pkg\"\n          rm -f <%~ it.vars.specFeedbackFile %> <%~ it.vars.nextFile %>\n        on:\n          \"* **\": picking\n\nuse:\n  # Shared health triple: checking/fixing/escalate -> reviewing.\n  - submachine: makeGreen\n    name: build\n    as: { check: checking, fix: fixing, escalate: escalate }\n    with:\n      onGreen: reviewing\n      escalateDescribe: >-\n        Fix the failure yourself — edit the code and/or `.gtd/FEEDBACK.md` —\n        to re-run the check (**checking**).\n      checkLabel: Running checks\n      fixLabel: Fixing the check\n      escalateLabel: Escalating to a human\n  # Advanced package health triple: adv-checking/adv-fixing/adv-escalate -> spec-review.\n  - submachine: makeGreen\n    name: pkg\n    as: { check: adv-checking, fix: adv-fixing, escalate: adv-escalate }\n    with:\n      onGreen: spec-review\n      escalateDescribe: >-\n        Fix the failure yourself — edit the code and/or\n        `.gtd/FEEDBACK.md` — to re-run the check (**adv-checking**).\n      checkLabel: Running checks\n      fixLabel: Fixing the check\n      escalateLabel: Escalating to a human\n\n  # Green-baseline gate: simple flow entry.\n  - submachine: assertGreen\n    name: start\n    as: { check: start-check, blocked: start-blocked }\n    with:\n      onGreen: planning\n      checkScript: |\n        #!/usr/bin/env bash\n        # gtd check turn — run the suite before starting new work; a red run\n        # records .gtd/FEEDBACK.md (halt), a green run cleans it up (proceed).\n        set +e\n        mkdir -p .gtd\n        # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n        # newline after an interpolation tag, so no tag may be the last token on\n        # a line — it would glue the next line's `else`/`fi` onto it and break\n        # the script. Below this point everything is plain bash.\n        feedback=\"<%~ it.vars.feedbackFile %>\"\n        <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n        code=$?\n        if [ \"$code\" -ne 0 ]; then\n          if [ -s .gtd/.check-output ]; then\n            mv .gtd/.check-output \"$feedback\"\n          else\n            rm -f .gtd/.check-output\n            printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n          fi\n        else\n          rm -f .gtd/.check-output\n          rm -f \"$feedback\"\n        fi\n      blockedMessage: |\n        The test baseline is red — gtd will not start new work on a broken suite.\n        `<%= it.vars.feedbackFile %>` holds the failing output.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      blockedDescribe: >-\n        Fix the failing tests — edit the code and/or `.gtd/FEEDBACK.md` — to\n        re-run the baseline check (**start-check**). To repair the baseline as\n        its own separate reviewed commit instead, abandon this start and run\n        `gtd fix` from a clean `idle`.\n      checkLabel: Checking the baseline\n      blockedLabel: Baseline is red\n\n  # Green-baseline gate: advanced flow entry.\n  - submachine: assertGreen\n    name: adv-start\n    as: { check: adv-start-check, blocked: adv-start-blocked }\n    with:\n      onGreen: adv-grilling\n      checkScript: |\n        #!/usr/bin/env bash\n        # gtd check turn — run the suite before starting the advanced flow; a red\n        # run records .gtd/FEEDBACK.md (halt), a green run cleans it up (proceed).\n        set +e\n        mkdir -p .gtd\n        # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n        # newline after an interpolation tag, so no tag may be the last token on\n        # a line — it would glue the next line's `else`/`fi` onto it and break\n        # the script. Below this point everything is plain bash.\n        feedback=\"<%~ it.vars.feedbackFile %>\"\n        <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n        code=$?\n        if [ \"$code\" -ne 0 ]; then\n          if [ -s .gtd/.check-output ]; then\n            mv .gtd/.check-output \"$feedback\"\n          else\n            rm -f .gtd/.check-output\n            printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n          fi\n        else\n          rm -f .gtd/.check-output\n          rm -f \"$feedback\"\n        fi\n      blockedMessage: |\n        The test baseline is red — gtd will not start new work on a broken suite.\n        `<%= it.vars.feedbackFile %>` holds the failing output.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      blockedDescribe: >-\n        Fix the failing tests — edit the code and/or `.gtd/FEEDBACK.md` — to\n        re-run the baseline check (**adv-start-check**). To repair the baseline\n        as its own separate reviewed commit instead, abandon this start and run\n        `gtd fix` from a clean `idle`.\n      checkLabel: Checking the baseline\n      blockedLabel: Baseline is red\n\n  # Green-baseline gate: `gtd review` entry (reviewEntry -> reviewing).\n  - submachine: assertGreen\n    name: review-start\n    as: { check: review-start-check, blocked: review-start-blocked }\n    set:\n      check: { reviewEntry: true }\n    with:\n      onGreen: reviewing\n      checkScript: |\n        #!/usr/bin/env bash\n        # gtd check turn — run the suite before reviewing; a red run records\n        # .gtd/FEEDBACK.md (halt), a green run cleans it up (proceed to reviewing).\n        set +e\n        mkdir -p .gtd\n        # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n        # newline after an interpolation tag, so no tag may be the last token on\n        # a line — it would glue the next line's `else`/`fi` onto it and break\n        # the script. Below this point everything is plain bash.\n        feedback=\"<%~ it.vars.feedbackFile %>\"\n        <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n        code=$?\n        if [ \"$code\" -ne 0 ]; then\n          if [ -s .gtd/.check-output ]; then\n            mv .gtd/.check-output \"$feedback\"\n          else\n            rm -f .gtd/.check-output\n            printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n          fi\n        else\n          rm -f .gtd/.check-output\n          rm -f \"$feedback\"\n        fi\n      blockedMessage: |\n        The test baseline is red — gtd will not start a review on a broken suite.\n        `<%= it.vars.feedbackFile %>` holds the failing output.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      blockedDescribe: >-\n        Fix the failing tests — edit the code and/or `.gtd/FEEDBACK.md` — to\n        re-run the baseline check (**review-start-check**).\n      checkLabel: Checking the baseline\n      blockedLabel: Baseline is red\n\n  # Advanced flow: product Q&A loop (adv-grilling/adv-grilling-answer -> architecting).\n  - submachine: qaLoop\n    name: grill\n    as: { author: adv-grilling, answer: adv-grilling-answer }\n    with:\n      authorFile: <%= it.vars.requirementsFile %>\n      onAdvance: architecting\n      authorPrompt: |\n        You are an autonomous coding agent. `.gtd/` holds this workflow's own\n        state — never create, edit, or delete anything under `.gtd/` except\n        `<%= it.vars.requirementsFile %>`, which this prompt tells you to develop.\n\n        `<%= it.vars.requirementsFile %>` holds the product requirements (or the\n        plan you already wrote plus the human's answers on a return lap). Read it,\n        explore the codebase, and develop it into a concrete, PRODUCT-level plan\n        in this one turn.\n\n        Scope: product and user-facing decisions only. Leave implementation\n        details (file/module structure, data models, tech-stack choices) for the\n        next phase (`<%= it.vars.architectureFile %>`).\n\n        For every open point you cannot settle yourself, add a question under a\n        `## Open Questions` heading near the top: one `### <question>` sub-heading\n        each, followed by a checkbox list of candidate answers — aim for TWO\n        concrete options plus a final free-text slot, ALL left UNTICKED:\n\n            ### <the question>\n\n            - [ ] <first option — a concrete answer, a few words of rationale>\n            - [ ] <second option>\n            - [ ] _your answer_\n\n        Never tick a box yourself — the human ticks exactly one per question. The\n        last `- [ ] _your answer_` line is the human's own-answer slot; write it\n        with the literal `_your answer_` placeholder and leave it for them.\n\n        On a RETURN lap the human has ticked one option per question (or replaced\n        `_your answer_` with their own text). For each answered question: fold the\n        chosen answer into the plan prose, then MOVE the question down into a\n        `## Answered Questions` section as `### <question>` followed by the\n        resolved answer in plain PROSE (drop the checkboxes). Never re-raise a\n        question the human deleted; if the human deleted the whole\n        `## Open Questions` section, treat the plan as accepted and surface no new\n        questions unless something is genuinely still blocking. Omit\n        `## Open Questions` entirely once there are none left.\n\n        Leave `<%= it.vars.requirementsFile %>` uncommitted and finish your turn.\n      answerMessage: |\n        `<%= it.vars.requirementsFile %>` holds the product plan under\n        development. Each question under `## Open Questions` offers a few options\n        plus a `- [ ] _your answer_` slot. Answer EVERY question by ticking\n        exactly one box (`- [x]`); for your own answer, replace `_your answer_`\n        with your text and tick that line. Stepping is refused while any question\n        is unanswered.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      advanceDescribe: >-\n        With no open questions left (the agent surfaced none, or you deleted\n        the section and it finalized the plan), change nothing to accept the\n        product plan and move on to technical architecting (**architecting**).\n      editDescribe: >-\n        Tick exactly one option per open question (replace `_your answer_` for\n        your own) to send the plan back for the agent to fold your answers in\n        (**adv-grilling**). Delete a question to skip it, or delete the whole\n        `## Open Questions` section to accept the plan as-is.\n      authorLabel: Refining the product plan\n      answerLabel: Awaiting your answers\n\n  # Advanced flow: technical Q&A loop (architecting/architecting-answer -> decompose).\n  - submachine: qaLoop\n    name: arch\n    as: { author: architecting, answer: architecting-answer }\n    with:\n      authorFile: <%= it.vars.architectureFile %>\n      onAdvance: decompose\n      authorPrompt: |\n        You are an autonomous coding agent. `.gtd/` holds this workflow's own\n        state — never create, edit, or delete anything under `.gtd/` except\n        `<%= it.vars.architectureFile %>` and `<%= it.vars.requirementsFile %>`,\n        which this prompt tells you to write and delete respectively.\n\n        Read `<%= it.vars.requirementsFile %>` (the converged product plan) and\n        develop `<%= it.vars.architectureFile %>` from it in this one turn:\n        file/module structure, data models, library/tech-stack choices,\n        error-handling strategy — the *how*, building on the *what* already\n        settled. Do not re-open product-level decisions; treat them as settled.\n\n        For every open point you cannot settle yourself, add a question under a\n        `## Open Questions` heading exactly as the grilling phase does: one\n        `### <question>` each, followed by a checkbox list — TWO concrete options\n        plus a final free-text slot, ALL left UNTICKED:\n\n            ### <the question>\n\n            - [ ] <first option>\n            - [ ] <second option>\n            - [ ] _your answer_\n\n        Never tick a box yourself. On a RETURN lap the human has ticked one option\n        per question (or replaced `_your answer_` with their own text): fold each\n        chosen answer into the plan prose, then MOVE the question into a\n        `## Answered Questions` section as `### <question>` + the resolved answer\n        in plain PROSE (drop the checkboxes). Never re-raise a deleted question;\n        treat a deleted `## Open Questions` section as acceptance. Omit\n        `## Open Questions` once there are none left.\n\n        Once `<%= it.vars.architectureFile %>` is written, delete\n        `<%= it.vars.requirementsFile %>` — its content has been folded in and it\n        must not linger. Leave everything uncommitted and finish your turn.\n      answerMessage: |\n        `<%= it.vars.architectureFile %>` holds the technical plan under\n        development. Each question under `## Open Questions` offers a few options\n        plus a `- [ ] _your answer_` slot. Answer EVERY question by ticking\n        exactly one box (`- [x]`); for your own answer, replace `_your answer_`\n        with your text and tick that line. Stepping is refused while any question\n        is unanswered.\n\n        What each change does next (then run `gtd step human`):\n        <% it.edges.forEach(function (e) { if (e.describe) { %>\n        <%~ \"- \" + e.describe + \"\\n\" %>\n        <% } }) %>\n      advanceDescribe: >-\n        With no open questions left (the agent surfaced none, or you deleted\n        the section and it finalized the plan), change nothing to accept the\n        technical plan and move on to decomposing the work into packages\n        (**decompose**).\n      editDescribe: >-\n        Tick exactly one option per open question (replace `_your answer_` for\n        your own) to send the plan back for the agent to fold your answers in\n        (**architecting**). Delete a question to skip it, or delete the whole\n        `## Open Questions` section to accept the plan as-is.\n      authorLabel: Refining the technical plan\n      answerLabel: Awaiting your answers\n\n  # Simple flow plan-iteration loop (planning/plan-review -> building).\n  - submachine: planLoop\n    with:\n      onAccept: building\n\n  # Shared review tail (reviewing -> await-review -> ... -> squashing).\n  - submachine: humanReview\n    with:\n      onSignoff: squashing\n\n  # Advanced per-package agentic review (spec-review/spec-fix -> closing).\n  - submachine: specGate\n    with:\n      onApproved: closing\n\n  # Advanced per-package build queue (picking/adv-building/closing -> reviewing).\n  - submachine: packageLoop\n    with:\n      onDrained: reviewing\n\nstates:\n  # ── Shared entry ────────────────────────────────────────────────────────\n  idle:\n    actor: human\n    initial: true\n    label: Idle\n    # Every human-gate `message:` below closes with the same rendered \"What\n    # each change does next\" list, built from `it.edges` (the resting state's\n    # own `on` edges — each edge's `describe:` is the single source of truth).\n    message: |\n      No active gtd cycle.\n\n      To start one, create a steering file describing what you want built:\n\n      - `.gtd/TODO.md` — a short sketch — starts the SIMPLE flow (plan, build,\n        review).\n      - `.gtd/REQUIREMENTS.md` — the product requirements — starts the ADVANCED\n        flow (product + technical Q&A, package decomposition, parallel build,\n        per-package agentic review).\n\n      What each change does next (then run `gtd step human`):\n      <% it.edges.forEach(function (e) { if (e.describe) { %>\n      <%~ \"- \" + e.describe + \"\\n\" %>\n      <% } }) %>\n    on:\n      \"* <%= it.vars.requirementsFile %>\":\n        to: adv-start-check\n        describe: >-\n          Save the product requirements to `.gtd/REQUIREMENTS.md` to start the\n          ADVANCED flow — gtd first checks the test baseline is green, then the\n          agent develops them through product then technical Q&A, decomposes the\n          work into packages, and builds each in parallel (**adv-start-check**).\n      \"* **\":\n        to: start-check\n        describe: >-\n          Save a short sketch to `.gtd/TODO.md` (or any other change) to start\n          the SIMPLE flow — gtd first checks the test baseline is green, then the\n          agent reads it, explores the codebase, and develops it into a concrete\n          implementation plan (**start-check**).\n\n  # ── Simple flow: monolithic build (plan loop is the `planLoop` sub-machine) ─\n  building:\n    actor: agent\n    label: Building\n    file: <%= it.vars.todoFile %>\n    model: <%= it.vars.coderModel %>\n    memory: build\n    prompt: |\n      You are an autonomous coding agent. `.gtd/` holds this workflow's own\n      state — never create or edit anything under `.gtd/` except deleting\n      `<%= it.vars.todoFile %>`, which this prompt tells you to delete once its\n      work is complete.\n\n      The plan to implement is: <%~ it.read(it.vars.todoFile) %>\n\n      Read that plan and implement it in full. Use TDD discipline: one test,\n      then the implementation that passes it, then the next — never write all\n      the tests first.\n\n      Delete `<%= it.vars.todoFile %>` once the work it describes is complete\n      and verified. Leave everything else uncommitted and finish your turn.\n    on:\n      \"* **\": checking\n\n  # ── Advanced flow: decomposition (Q&A loops are the `qaLoop` sub-machine) ───\n  decompose:\n    actor: agent\n    label: Decomposing into packages\n    model: <%= it.vars.coderModel %>\n    memory: build\n    prompt: |\n      You are an autonomous coding agent. `.gtd/` holds this workflow's own\n      state — never create, edit, or delete anything under `.gtd/` except the\n      package files this prompt tells you to write (under\n      `<%= it.vars.packagesDir %>/`), and `<%= it.vars.architectureFile %>`,\n      which it tells you to delete.\n\n      Read `<%= it.vars.architectureFile %>` (the converged technical plan) and\n      decompose it into an ordered set of WORK PACKAGES, one file each under\n      `<%= it.vars.packagesDir %>/` (e.g. `<%= it.vars.packagesDir %>/01-name.md`,\n      `<%= it.vars.packagesDir %>/02-name.md`, ...). Order the packages so the\n      test suite can pass again after each one lands.\n\n      Each package file describes a set of INDEPENDENT tasks that can be\n      implemented concurrently without ordering constraints between them — a\n      single build turn will fan them out to parallel subagents. Give each task\n      a clear description, acceptance criteria as `- [ ]` checkboxes, and the\n      relevant paths. Also state the package's overall requirement so it can be\n      reviewed against its spec. Do NOT put ordering-dependent tasks in the same\n      package — split them across ordered packages instead.\n\n      No package file may reference any other `.gtd/` file. Once the package\n      files are written, delete `<%= it.vars.architectureFile %>`. Leave\n      everything uncommitted and finish your turn.\n    on:\n      \"* <%= it.vars.packagesDir %>/**\": picking\n\n  # ── Green-baseline gate: `gtd fix` entry ──────────────────────────────────\n  # `gtd fix` enters HERE (fixEntry): run the suite, then a red run drops\n  # straight into the shared `fixing` loop (-> checking -> review + squash\n  # tail); a green run is a no-op back to `idle` (nothing to fix).\n  fix-check:\n    actor: check\n    fixEntry: true\n    label: Checking the baseline\n    script: |\n      #!/usr/bin/env bash\n      # gtd check turn — run the suite to find what needs fixing; a red run\n      # records .gtd/FEEDBACK.md (-> fixing), a green run cleans it up (-> idle).\n      set +e\n      mkdir -p .gtd\n      # Hoist the feedback path once, at the TOP: Eta's autoTrim eats the\n      # newline after an interpolation tag, so no tag may be the last token on\n      # a line — it would glue the next line's `else`/`fi` onto it and break\n      # the script. Below this point everything is plain bash.\n      feedback=\"<%~ it.vars.feedbackFile %>\"\n      <%~ it.vars.testCommand %> > .gtd/.check-output 2>&1\n      code=$?\n      if [ \"$code\" -ne 0 ]; then\n        if [ -s .gtd/.check-output ]; then\n          mv .gtd/.check-output \"$feedback\"\n        else\n          rm -f .gtd/.check-output\n          printf 'the test command failed with exit code %s and produced no output.' \"$code\" > \"$feedback\"\n        fi\n      else\n        rm -f .gtd/.check-output\n        rm -f \"$feedback\"\n      fi\n    on:\n      \"A <%= it.vars.feedbackFile %>\": fixing\n      \"M <%= it.vars.feedbackFile %>\": fixing\n      \"D <%= it.vars.feedbackFile %>\": idle\n      \"C\": idle\n\n  # ── Shared tail: squash finale (review loop is the `humanReview` sub-machine) ─\n  squashing:\n    actor: agent\n    label: Squashing\n    file: <%= it.vars.commitMsgFile %>\n    model: <%= it.vars.coderModel %>\n    memory: build\n    prompt: |\n      You are an autonomous coding agent. The cycle is approved and done.\n      `.gtd/` holds this workflow's own state — never create, edit, or delete\n      anything under `.gtd/` except `<%= it.vars.commitMsgFile %>`, which this\n      prompt tells you to write.\n\n      Leave `<%= it.vars.commitMsgFile %>` uncommitted and finish your turn —\n      entering this file squashes everything this process retains into one\n      commit using its content as the message. The diff below is exactly what\n      that commit will contain: `it.retainedDiff`, the changes THIS process\n      made (its trace/retry boundary onward). For a normal build cycle that is\n      the whole feature; for a `gtd review` process it is only the fixes made\n      during the review — NOT the changeset that was under review.\n      <% if (it.retainedDiff.trim()) { %>\n\n      Write `<%= it.vars.commitMsgFile %>` with ONE conventional-commits message\n      for the changes below (subject line `type(scope): subject`, imperative\n      mood, <= 72 characters; a body explaining the why, trade-offs, and key\n      decisions). Plain text, no markdown wrapper.\n\n      ## Retained diff\n\n      ```diff\n      <%~ it.retainedDiff %>\n      ```\n      <% } else { %>\n\n      This process retained NO changes (a clean review sign-off with no fixes).\n      Write `<%= it.vars.commitMsgFile %>` with exactly this one line and\n      nothing else:\n\n      chore: human review\n      <% } %>\n    on:\n      \"A <%= it.vars.commitMsgFile %>\": done\n      \"M <%= it.vars.commitMsgFile %>\": done\n\n  done:\n    commit: |\n      <%~ it.read(it.vars.commitMsgFile) %>\n\n      Token cost: <%= it.processCost %>\n      <% it.processCostByModel.forEach(function(m){ %>\n      - <%= m.model %>: <%= m.cost %>\n      <% }) %>\n";
 //#endregion
 //#region src/workflows/templates.ts
 /**

package/docs/cli.md

@@ -48,7 +48,9 @@ Commands:
   edit [path]      Open <path> (repo-relative) in $VISUAL/$EDITOR, blocking
                    until the editor exits. With no argument, opens the
                    resolved rest's declared file (or the repo root if it
-                   declares none). Never reads --no-edit/GTD_NO_EDIT
+                   declares none). Never reads --no-edit/--edit/GTD_NO_EDIT.
+                   Low-level plumbing — --edit covers the common "force an
+                   edit, then keep driving" case
   status           Print the resolved rest's state/actor and which declared
                    pattern (if any) each pending change matches (no mutation)
   validate         Format and validate the steering file the resolved rest
@@ -69,6 +71,13 @@ Options:
   --model=<name>   (gtd step only, with --cost) tag that cost's model
   --no-edit        (bare gtd or gtd loop only) disable the loop's automatic
                    editor launching at human gates
+  --edit, -e       (bare gtd or gtd loop only) force the editor open at the
+                   current human gate, overriding --no-edit/GTD_NO_EDIT; a
+                   no-op note (not an error) when not at a human gate
+  --once           (bare gtd or gtd loop only) run exactly one loop beat (one
+                   human gate, one script check+step, or one agent
+                   prompt+step), then exit — combines freely with
+                   --edit/--no-edit
   --version, -v    Print version and exit
   --help, -h       Print this help and exit
 ```
@@ -92,25 +101,52 @@ to plain-text mode. A bare `--cost`/`--model` with no value, a non-numeric or
 negative `--cost`, an empty `--model`, `--model` without `--cost`, or either
 flag on any command other than `gtd step` are all usage errors.
 
-`--no-edit` is a separate, bash-level flag handled entirely by `bin/gtd` itself,
-stripped before anything reaches the bundle — see [`--no-edit`](#--no-edit)
+`--no-edit`, `--edit`/`-e`, and `--once` are separate, bash-level flags handled
+entirely by `bin/gtd` itself, stripped before anything reaches the bundle — see
 below.
 
-## `--no-edit`
-
-A flag on the loop driver only — bare `gtd --no-edit` or `gtd loop --no-edit`
-(the two are equivalent, since bare `gtd` and `gtd loop` are themselves
-equivalent). Recognized in exactly those two positions; any other
-placement/combination (e.g. `gtd step --no-edit`, or combined with any other
-argument) is a usage error. The `GTD_NO_EDIT` environment variable (any
-non-empty value) does the same thing.
-
-Disables the loop's default behavior of launching an editor at human gates,
-restoring the previous halt-and-print-and-exit behavior with no editor involved.
-It governs only the loop's own automatic launching — `gtd edit` invoked directly
-never reads it (see [`gtd edit`](#gtd-edit-path) above). See
-[Driving the loop](loop.md) and `skills/loop/SKILL.md` for the full gate-flow
-description.
+## `--no-edit`, `--edit`, `--once`
+
+Three flags on the loop driver only (`--edit` also has the short form `-e`),
+each recognized bare (`gtd --no-edit`) or immediately after `loop`
+(`gtd loop --no-edit`) — the two positions are equivalent, since bare `gtd` and
+`gtd loop` are themselves equivalent. All three combine freely with each other,
+in any order (e.g. `gtd --edit --once`, `gtd loop --once --edit`), except
+`--edit`/`--no-edit` together, which conflict (forcing and suppressing the
+editor in the same run is not a coherent request). Any other placement (e.g.
+`gtd step --once`), a duplicated flag (`gtd --once --once`), or the
+`--edit`/`--no-edit` conflict is a usage error, surfaced via whatever guard the
+misplaced argument first reaches (the compiled bundle's own
+unknown-option/unknown-command rejection for a misplaced flag, or
+`gtd: 'loop' takes no arguments` for one misplaced after `loop`) — never
+silently ignored or silently resolved one way.
+
+**`--no-edit`** (or the `GTD_NO_EDIT` environment variable, any non-empty value)
+disables the loop's default behavior of launching an editor at human gates,
+restoring the halt-and-print-and-exit behavior with no editor involved.
+
+**`--edit`** (`-e`) is `--no-edit`'s mirror: it forces the editor open at the
+current human gate, overriding an ambient `GTD_NO_EDIT`/`--no-edit` default for
+this one run. It is only meaningful when the machine currently rests at a human
+gate — the honesty caveat: at any other rest, forcing an edit is undefined, so
+`bin/gtd` prints a note that it isn't at a human gate yet and simply keeps
+driving, rather than pretending to force something. Both flags govern only the
+loop's own automatic launching — `gtd edit` invoked directly never reads either
+(see [`gtd edit`](#gtd-edit-path) above).
+
+**`--once`** restricts the loop to exactly one beat — one human gate, one script
+check+step, or one agent prompt+step — then exits, instead of driving all the
+way back to idle/settled. Precisely: at most one commit/transition is made;
+whichever of the three kinds the currently-resolved rest is, that one gets
+processed (including any internal `gtd validate` fix-reprompt rounds a producing
+agent turn needs — those are all part of finishing the ONE prompt beat, not
+separate beats) and the loop exits 0 immediately after, without re-peeking at
+what comes next. A clean human gate with nothing to capture (the opening move's
+silent capture attempt under `--no-edit`, or a no-op editor session) is not a
+beat and falls through to the driver's ordinary halt, same as without `--once`.
+
+See [Driving the loop](loop.md) and `skills/loop/SKILL.md` for the full
+gate-flow description.
 
 ## `gtd init`
 
@@ -436,6 +472,12 @@ which are JSON-only. `--json` emits
 
 ## `gtd edit [path]`
 
+Low-level plumbing: for the common case of "force an edit at the current human
+gate, then keep driving the loop", use `--edit` (below) instead. `gtd edit`
+itself only opens an editor on a path — it never drives anything afterward,
+which makes it the right tool for opening a SPECIFIC path (or peeking at the
+current rest's file) without starting a loop.
+
 Handled entirely in `bin/gtd`'s own bash — unlike every other subcommand, it is
 **never forwarded** to the compiled bundle (`dist/gtd.bundle.mjs`).
 
@@ -457,8 +499,8 @@ gtd edit: could not determine the next step:
 <gtd next --json's own error output>
 ```
 
-`gtd edit` **never** reads `--no-edit`/`GTD_NO_EDIT` — those two govern only the
-loop's own automatic editor launching at human gates (see
+`gtd edit` **never** reads `--no-edit`/`--edit`/`GTD_NO_EDIT` — those govern
+only the loop's own automatic editor launching at human gates (see
 [Driving the loop](loop.md) and `skills/loop/SKILL.md`), not this command.
 Invoked directly, `gtd edit` always launches, unconditionally.
 

package/docs/loop.md

@@ -66,11 +66,25 @@ the state declares none), blocking until the editor exits, then runs
   exactly as a manual `gtd step human` refusal would.
 
 Set `GTD_NO_EDIT` (any non-empty value), or pass `--no-edit` — see
-[`cli.md`'s `--no-edit`](cli.md#--no-edit) for its exact syntax and positioning
+[`cli.md`](cli.md#--no-edit---edit---once) for its exact syntax and positioning
 — to disable this and restore the original halt-and-print behavior: the loop
 prints the gate via `gtd next` and exits 0 without touching an editor, leaving
 the human to edit and re-launch the loop themselves.
 
+Pass `--edit` (`-e`) to force the editor open at the current human gate right
+now, overriding an ambient `GTD_NO_EDIT`/`--no-edit` for this one run — useful
+when a shell profile sets `GTD_NO_EDIT` by default but you want to edit
+interactively just this once. It only means something when the machine is
+actually resting at a human gate: at any other rest, `bin/gtd` prints a note
+that it isn't at one yet and simply keeps driving, rather than pretending to
+force something undefined.
+
+Pass `--once` to restrict the loop to exactly one beat — one human gate, one
+script check+step, or one agent prompt+step — then exit, instead of driving all
+the way to idle/settled. It combines freely with `--edit`/`--no-edit` (e.g.
+`gtd --edit --once` forces the editor at the current gate, captures the edit,
+and stops right there rather than continuing to drive).
+
 ```bash
 gtd next --json   # ask who's up and what they should do
 ```
@@ -92,18 +106,21 @@ when the machine rests at a `"message"` gate — so `gtd` is the only command a
 human runs. `bin/gtd` is the packaged entry point: invoked bare, or with `loop`
 as its first argument, it runs this exact script; any other first argument (e.g.
 `next`, `step`, `status`) hands off to `node dist/gtd.bundle.mjs` instead. The
-loop body it runs adds five things on top of the reference script above: it
-opens an editor at every `"message"` gate instead of just printing it (see
-"Editing is on by default" above; `--no-edit`/`GTD_NO_EDIT` fall back to the
-reference script's plain print-and-exit); it stops with a diagnostic if the same
-`"prompt"` state/content repeat with no progress (see `skills/loop/SKILL.md`'s
-"Stall detection"); it lets `GTD_LOOP_AGENT_CMD` swap in any coding agent CLI in
-place of the default `claude -p`, receiving the prompt via `$GTD_LOOP_PROMPT`;
-it exports the resolved state's optional `model` hint as `$GTD_LOOP_MODEL`,
-appending `--model "$GTD_LOOP_MODEL"` to the default `claude -p` invocation
-whenever it's non-empty; and it acts on the optional `memory` scope hint (see
-"Agent memory scope" below) — continuing one agent session across consecutive
-same-scope turns and starting fresh when the scope changes.
+loop body it runs adds six things on top of the reference script above: it opens
+an editor at every `"message"` gate instead of just printing it (see "Editing is
+on by default" above; `--no-edit`/`GTD_NO_EDIT` fall back to the reference
+script's plain print-and-exit, while `--edit`/`-e` forces it on for this run
+even over an ambient `GTD_NO_EDIT`); it can restrict itself to exactly one beat
+via `--once` instead of driving to idle/settled; it stops with a diagnostic if
+the same `"prompt"` state/content repeat with no progress (see
+`skills/loop/SKILL.md`'s "Stall detection"); it lets `GTD_LOOP_AGENT_CMD` swap
+in any coding agent CLI in place of the default `claude -p`, receiving the
+prompt via `$GTD_LOOP_PROMPT`; it exports the resolved state's optional `model`
+hint as `$GTD_LOOP_MODEL`, appending `--model "$GTD_LOOP_MODEL"` to the default
+`claude -p` invocation whenever it's non-empty; and it acts on the optional
+`memory` scope hint (see "Agent memory scope" below) — continuing one agent
+session across consecutive same-scope turns and starting fresh when the scope
+changes.
 
 `bin/gtd` resolves both its bundle hand-off and the loop's own internal `gtd`
 calls against `dist/gtd.bundle.mjs` next to its own location. Set `GTD_BIN` — a

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pmelab/gtd",
-  "version": "7.6.2",
+  "version": "7.7.0",
   "private": false,
   "description": "Git-aware CLI that emits the next prompt for an autonomous coding agent based on the current repository state",
   "bin": {