BaseInfinity · BaseInfinity · Jun 1, 2026 · Jun 1, 2026
diff --git a/ROADMAP.md b/ROADMAP.md
@@ -13,6 +13,18 @@ Anything else goes to the **Research Parking Lot** (below) with a named revisit
 
 Each row's status should be one of: **Actionable now** / **Waiting on trigger** / **Tracked elsewhere** / **DONE** / **KILLED**.
 
+## Research Parking Lot
+
+Speculative ideas that didn't qualify for entry under the Demand-Signal-First gate. Each entry MUST declare a **named revisit trigger** and an **expiry date** (30 days for implementation ideas, 60 days for research/watch items). If the trigger has not fired by expiry, the row is **deleted, not carried forward** — the parking lot is not a graveyard, it's a finite waiting room.
+
+Add entries when a candidate is mid-evidence (one signal exists but doesn't yet clear the gate's "second external user" or "dated platform deadline" thresholds). Promote to a numbered ROADMAP row only when the trigger fires.
+
+| Idea | Why parked | Revisit trigger | Expiry | Source |
+|---|---|---|---|---|
+| _(empty — populate as triage demands)_ | | | | |
+
+> **Maintenance rule:** during quarterly ROADMAP triage, prune any row past expiry. Logged removals go in the commit message (`docs(roadmap): prune parking lot — <N> expired entries`) so deletions are traceable.
+
 ## Next Up (v1.37.0 queue, priority-ordered 2026-04-23)
 
 0. **#212 — Local-Max E2E shepherd (zero-API)** — ✅ **ENGINEERING COMPLETE; PROVE-IT GATE DEFERRED 2026-05-05** ("just defer prove it who cares" — maintainer pragmatic call). Sim/eval/orchestration all on Max subscription per #228 (v1.59.0) + #230 + #231 phases 1-4 (v1.50.0–v1.55.0). Weekly+monthly workflow porting deleted ~$25-55/week of API cron burn. Docs reference `tests/e2e/local-shepherd.sh --compare-baseline` as the default path (`CLAUDE_CODE_SDLC_WIZARD.md` lines 42, 102, 1325, 2570). **Codex cross-model review** runs against ChatGPT/Codex subscription auth (`codex login`, "Logged in using ChatGPT" 2026-05-05) — also $0 marginal. Net: **$0/release** (was $20+/release in API burn). **Prove-It Gate parked** — would require ~$20 one-time API spend to run paired N=15 local-Max vs CI-API trials for 95% CI overlap proof. Not worth the spend; engineering signal works fine in practice. Reopen if local scoring drifts vs known-good baselines. Original scope archived below for history. <details><summary>Original scope (archived)</summary>**TOP PRIORITY** after 2026-04-23 live-fire proved the pain. Today's shepherding blew through the Anthropic API credit cap mid-release (PR #222 e2e-quick-check failed with "Credit balance is too low"), blocking v1.36.1 until admin-merge. Current CI spends ~$0.62 (Tier 1) + $0.82 (Tier 2) per PR on `anthropics/claude-code-action@v1`. Across today's 12 PRs that's **$20+ in API burn** just for simulations. **Architecture** (tri-split billing): (a) **Simulation** — `claude --print` via local Bash tool runs against **Max subscription** ($0 after monthly cap, tests wizard behavior ON Claude which is the whole point); (b) **Cross-model review of PR diff** — `codex exec -c model_reasoning_effort=xhigh` via local Bash tool uses **OpenAI API** (separate bill, ~$3-5/PR — REVISED 2026-05-05: $0 if Codex authed via ChatGPT subscription); (c) **Orchestration** (watching CI, pulling logs, posting PR comments, merging) — me running in local terminal on **Max quota** ($0). **Scope:** (1) replace `claude-code-action@v1` invocation in `run-simulation.sh` with `claude --print "$prompt" --output-format json`; (2) confirm execution-output JSON shape matches (SDK format); (3) keep CI path as fallback for external contributors without a Max sub; (4) new `./tests/e2e/local-shepherd.sh <PR>` orchestrator runs Tier 1+2 locally, posts summary; (5) update SDLC skill + CLAUDE_CODE_SDLC_WIZARD.md docs to reference local shepherd as the default path. **Prove-It Gate** (CRITICAL — unchanged): demonstrate statistical score parity (local-Max vs CI-API) on **at least 3 PRs** before trusting local signal. Both paths hit same model (Opus 4.7), same prompt, same scoring — expected differences are only stochastic variance (±1-2 pts typical), so prove via overlapping 95% CIs, not byte-equality. If parity fails, investigate whether Max routing differs from raw-API before declaring. User call-out 2026-04-23: "why are we using API credits for CI instead of us being CI shepherd?"</details>

diff --git a/cli/init.js b/cli/init.js
@@ -25,6 +25,7 @@ const FILES = [
   { src: 'hooks/instructions-loaded-check.sh', dest: '.claude/hooks/instructions-loaded-check.sh', executable: true, base: REPO_ROOT },
   { src: 'hooks/model-effort-check.sh', dest: '.claude/hooks/model-effort-check.sh', executable: true, base: REPO_ROOT },
   { src: 'hooks/precompact-seam-check.sh', dest: '.claude/hooks/precompact-seam-check.sh', executable: true, base: REPO_ROOT },
+  { src: 'hooks/goal-confidence-check.sh', dest: '.claude/hooks/goal-confidence-check.sh', executable: true, base: REPO_ROOT },
   // #254 Bug 1: shared helper sourced by all hooks above. Must ship — without
   // it, hooks emit "_find-sdlc-root.sh: No such file or directory" + the
   // SDLC root walk-up logic is silently dead.

diff --git a/cli/templates/settings.json b/cli/templates/settings.json
@@ -7,6 +7,10 @@
           {
             "type": "command",
             "command": "\"$CLAUDE_PROJECT_DIR\"/.claude/hooks/sdlc-prompt-check.sh"
+          },
+          {
+            "type": "command",
+            "command": "\"$CLAUDE_PROJECT_DIR\"/.claude/hooks/goal-confidence-check.sh"
           }
         ]
       }

diff --git a/hooks/goal-confidence-check.sh b/hooks/goal-confidence-check.sh
@@ -0,0 +1,82 @@
+#!/bin/bash
+# /goal SDLC discipline gate (ROADMAP #360).
+#
+# UserPromptSubmit hook. When the user invokes `/goal <condition>`, scan the
+# prior assistant text for a HIGH 95% confidence statement AND check the
+# condition for a DLC binding (`/sdlc`, `/gdlc`, `/ldlc`, etc.). Emit loud
+# warnings on either gap so the SDLC-discipline gap surfaces BEFORE the goal
+# evaluator starts rubber-stamping flailing as progress.
+#
+# PR #355 (v1.77.0) added the guidance text in skills/sdlc/SKILL.md. This
+# hook is the enforcement layer per #360. Non-blocking soft nudge — same
+# pattern as model-effort-check.sh.
+
+HOOK_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+source "$HOOK_DIR/_find-sdlc-root.sh"
+
+# Plugin + project copies both register this hook → plugin yields (#236).
+dedupe_plugin_or_project "${BASH_SOURCE[0]}" || exit 0
+
+# Must be invoked via hook (stdin JSON), not interactively.
+[ -t 0 ] && exit 0
+STDIN_JSON=$(cat)
+[ -z "$STDIN_JSON" ] && exit 0
+command -v jq > /dev/null 2>&1 || exit 0
+
+PROMPT=$(printf '%s' "$STDIN_JSON" | jq -r '.prompt // empty' 2>/dev/null) || exit 0
+[ -z "$PROMPT" ] && exit 0
+
+# Only act on `/goal <condition>` invocations. `/goal` alone (status) and
+# `/goal clear` (reset) must stay silent — those aren't discipline events.
+case "$PROMPT" in
+    "/goal")           exit 0 ;;
+    "/goal "[[:space:]]*) exit 0 ;;
+    "/goal clear")     exit 0 ;;
+    "/goal clear "*)   exit 0 ;;
+    "/goal "*)         ;;
+    *)                 exit 0 ;;
+esac
+
+CONDITION="${PROMPT#/goal }"
+
+# === Confidence gate ===
+# Read transcript_path, slurp JSONL, find last assistant message with text
+# content, scan for HIGH-95% confidence pattern. Silent on missing/unreadable
+# transcript — better to under-warn than crash a soft nudge.
+TRANSCRIPT_PATH=$(printf '%s' "$STDIN_JSON" | jq -r '.transcript_path // empty' 2>/dev/null) || TRANSCRIPT_PATH=""
+HAS_CONFIDENCE=0
+if [ -n "$TRANSCRIPT_PATH" ] && [ -r "$TRANSCRIPT_PATH" ]; then
+    LAST_ASSISTANT_TEXT=$(jq -rs '
+        [.[] | select(.type == "assistant")]
+        | last
+        | if . == null then ""
+          else (.message.content // []) | map(select(.type == "text") | .text) | join("\n")
+          end
+    ' "$TRANSCRIPT_PATH" 2>/dev/null) || LAST_ASSISTANT_TEXT=""
+    if printf '%s' "$LAST_ASSISTANT_TEXT" | grep -qiE 'HIGH \(?95%|confidence:?[[:space:]]+HIGH|HIGH[[:space:]]+95%|95%[[:space:]]+confidence'; then
+        HAS_CONFIDENCE=1
+    fi
+fi
+
+if [ "$HAS_CONFIDENCE" -eq 0 ]; then
+    echo ""
+    echo "!! /GOAL CONFIDENCE GATE: prior turn did not state HIGH 95% confidence !!"
+    echo "   The /goal Haiku evaluator rubber-stamps flailing as progress below 95%."
+    echo "   Recommendation: /goal clear → do research → state HIGH 95% confidence → re-issue /goal."
+    echo "   (Source: skills/sdlc/SKILL.md Long-Running Goals section, ROADMAP #360.)"
+    echo ""
+fi
+
+# === DLC binding gate ===
+# Condition MUST name a DLC (`/sdlc`, `/gdlc`, `/ldlc`, etc.) so the
+# evaluator anchors on "doing it right" not just "doing it".
+if ! printf '%s' "$CONDITION" | grep -qiE '/[a-z]+dlc'; then
+    echo ""
+    echo "!! /GOAL DLC-BINDING GATE: condition does not name a DLC (/sdlc, /gdlc, /ldlc) !!"
+    echo "   The evaluator needs a discipline anchor to judge 'doing it right' vs flailing."
+    echo "   Recommendation: rewrite condition to include 'following /sdlc' (or /gdlc/etc)."
+    echo "   (Source: skills/sdlc/SKILL.md Long-Running Goals section, ROADMAP #360.)"
+    echo ""
+fi
+
+exit 0
diff --git a/hooks/hooks.json b/hooks/hooks.json
@@ -7,6 +7,10 @@
           {
             "type": "command",
             "command": "${CLAUDE_PLUGIN_ROOT}/hooks/sdlc-prompt-check.sh"
+          },
+          {
+            "type": "command",
+            "command": "${CLAUDE_PLUGIN_ROOT}/hooks/goal-confidence-check.sh"
           }
         ]
       }

diff --git a/skills/sdlc/SKILL.md b/skills/sdlc/SKILL.md
@@ -141,6 +141,8 @@ PROTOCOL is universal across domains; only `review_instructions` and `verificati
 
 **Convergence:** 2 rounds sweet spot, 3 max. After 3 NOT CERTIFIED → escalate.
 
+**Enforcement:** `hooks/goal-confidence-check.sh` warns when `/goal` skips the 95%-confidence or DLC-binding gates (#360).
+
 **Multi-reviewer:** respond to each reviewer independently (no shared anchoring). **Non-code domains:** add `"audience"`/`"stakes"` keys to handoff.
 
 ### Release Review Focus

diff --git a/tests/test-hooks.sh b/tests/test-hooks.sh
@@ -3383,6 +3383,99 @@ test_hook_emits_cc_nudge_when_pending
 test_hook_silent_when_no_pending_cc_updates
 test_hook_silent_without_weekly_update_workflow
 
+# ---- /goal confidence + DLC-binding gate (ROADMAP #360) ----
+# UserPromptSubmit hook fires on `/goal <condition>` invocations, reads
+# transcript_path, scans the LAST assistant text message for a HIGH-95%
+# confidence statement, and also checks the condition for DLC binding
+# (`/sdlc`, `/gdlc`, `/ldlc`, etc). PR #355 (v1.77.0) added the *guidance*
+# in skills/sdlc/SKILL.md but nothing enforces it at runtime — this hook
+# is the enforcement layer per #360. Non-blocking soft nudge pattern
+# (mirrors model-effort-check.sh).
+
+test_goal_confidence_check_fires_when_no_confidence() {
+    local tmpdir
+    tmpdir=$(mktemp -d)
+    echo '<!-- SDLC Wizard Version: 1.77.0 -->' > "$tmpdir/SDLC.md"
+    touch "$tmpdir/TESTING.md"
+    cat > "$tmpdir/transcript.jsonl" <<'JSONL'
+{"type":"user","message":{"role":"user","content":"hey"}}
+{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"Working on this now."}]}}
+JSONL
+    local payload='{"prompt":"/goal \"ship X, following /sdlc, stop after 10 turns\"","transcript_path":"'"$tmpdir/transcript.jsonl"'"}'
+    local output
+    output=$(printf '%s' "$payload" | (cd "$tmpdir" && CLAUDE_PROJECT_DIR="$tmpdir" SDLC_WIZARD_CACHE_DIR="$tmpdir/cache" "$HOOKS_DIR/goal-confidence-check.sh"))
+    rm -rf "$tmpdir"
+    if echo "$output" | grep -qiE 'GATE|WARNING' && echo "$output" | grep -qi 'confidence'; then
+        pass "#360: /goal without prior HIGH 95% confidence triggers warning"
+    else
+        fail "#360: confidence-gate warning missing (got: $output)"
+    fi
+}
+
+test_goal_confidence_check_silent_when_confidence_present() {
+    local tmpdir
+    tmpdir=$(mktemp -d)
+    echo '<!-- SDLC Wizard Version: 1.77.0 -->' > "$tmpdir/SDLC.md"
+    touch "$tmpdir/TESTING.md"
+    cat > "$tmpdir/transcript.jsonl" <<'JSONL'
+{"type":"user","message":{"role":"user","content":"recon done?"}}
+{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"Confidence: HIGH (95%) — verified pattern matches existing reference."}]}}
+JSONL
+    local payload='{"prompt":"/goal \"ship X, following /sdlc, stop after 10 turns\"","transcript_path":"'"$tmpdir/transcript.jsonl"'"}'
+    local output
+    output=$(printf '%s' "$payload" | (cd "$tmpdir" && CLAUDE_PROJECT_DIR="$tmpdir" SDLC_WIZARD_CACHE_DIR="$tmpdir/cache" "$HOOKS_DIR/goal-confidence-check.sh"))
+    rm -rf "$tmpdir"
+    if echo "$output" | grep -qiE 'GATE|WARNING'; then
+        fail "#360: should be silent when prior turn has HIGH 95% confidence (got: $output)"
+    else
+        pass "#360: silent when prior turn has HIGH 95% confidence statement"
+    fi
+}
+
+test_goal_confidence_check_dlc_binding_warning() {
+    local tmpdir
+    tmpdir=$(mktemp -d)
+    echo '<!-- SDLC Wizard Version: 1.77.0 -->' > "$tmpdir/SDLC.md"
+    touch "$tmpdir/TESTING.md"
+    cat > "$tmpdir/transcript.jsonl" <<'JSONL'
+{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"Confidence: HIGH (95%)."}]}}
+JSONL
+    # Goal condition WITHOUT any DLC binding (/sdlc, /gdlc, /ldlc)
+    local payload='{"prompt":"/goal \"ship X, stop after 10 turns\"","transcript_path":"'"$tmpdir/transcript.jsonl"'"}'
+    local output
+    output=$(printf '%s' "$payload" | (cd "$tmpdir" && CLAUDE_PROJECT_DIR="$tmpdir" SDLC_WIZARD_CACHE_DIR="$tmpdir/cache" "$HOOKS_DIR/goal-confidence-check.sh"))
+    rm -rf "$tmpdir"
+    if echo "$output" | grep -qiE 'DLC.*binding|name.*DLC|/sdlc|/gdlc'; then
+        pass "#360: /goal without DLC binding triggers DLC-binding warning"
+    else
+        fail "#360: DLC-binding warning missing (got: $output)"
+    fi
+}
+
+test_goal_confidence_check_silent_on_status_and_clear() {
+    local tmpdir
+    tmpdir=$(mktemp -d)
+    echo '<!-- SDLC Wizard Version: 1.77.0 -->' > "$tmpdir/SDLC.md"
+    touch "$tmpdir/TESTING.md"
+    cat > "$tmpdir/transcript.jsonl" <<'JSONL'
+{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"some text"}]}}
+JSONL
+    local out1 out2
+    out1=$(printf '%s' '{"prompt":"/goal","transcript_path":"'"$tmpdir/transcript.jsonl"'"}' | (cd "$tmpdir" && CLAUDE_PROJECT_DIR="$tmpdir" SDLC_WIZARD_CACHE_DIR="$tmpdir/cache" "$HOOKS_DIR/goal-confidence-check.sh"))
+    out2=$(printf '%s' '{"prompt":"/goal clear","transcript_path":"'"$tmpdir/transcript.jsonl"'"}' | (cd "$tmpdir" && CLAUDE_PROJECT_DIR="$tmpdir" SDLC_WIZARD_CACHE_DIR="$tmpdir/cache" "$HOOKS_DIR/goal-confidence-check.sh"))
+    rm -rf "$tmpdir"
+    if [ -z "$out1" ] && [ -z "$out2" ]; then
+        pass "#360: silent on bare /goal (status) and /goal clear (reset)"
+    else
+        fail "#360: should be silent on status/clear (out1='$out1', out2='$out2')"
+    fi
+}
+
+test_goal_confidence_check_fires_when_no_confidence
+test_goal_confidence_check_silent_when_confidence_present
+test_goal_confidence_check_dlc_binding_warning
+test_goal_confidence_check_silent_on_status_and_clear
+
 echo ""
 echo "=== Results ==="
 echo "Passed: $PASSED"