@gobing-ai/spur 0.3.95 → 0.3.96

package/config/workflows/idea-pipeline.yaml

@@ -9,9 +9,9 @@
 #
 # Shape: start -> discovery -> idea-eval -> feature-create -> ac-generate -> feature-check
 #          -> system-design (conditional: needs_design signal)
-#          -> design-approval (taste HITL gate; not auto-clicked by --auto)
+#          -> design-approval (taste HITL gate; bypassed only when design_approved=true + design check PASS)
 #          -> decompose -> batch-create -> handoff-finalize -> handoff
-#        (idea-eval rejection -> cancelled)
+#        (idea-eval rejection, or auto + `drop` recommendation -> cancelled)
 #        (feature-check failure routes back to ac-generate; batch-create failure to decompose).
 #
 # Evidence instead of ceremony (task 0769): the feature check is MEASURED exactly once at
@@ -27,11 +27,11 @@
 #   design   — "auto" (default, signal-driven) or "skip" (--skip-design). No force path.
 #              Unified: skip also means omit per-task Design in batch-create (scaffold only;
 #              refine fills later). Default: decompose must author batch item `design`.
-#   design_approved — "true" when taste pre-cleared (CLI --approve-taste or alias --design-approved);
-#              lets profile=auto route around the design-approval taste gate (default "false")
-#   idea_approved — "true" when taste pre-cleared (CLI --approve-taste or alias --idea-approved);
-#              lets profile=auto route around the idea-eval taste gate (default "false")
-#   CLI --approve-taste sets both design_approved and idea_approved to true.
+#   design_approved — "true" lets profile=auto route around the design-approval taste gate when
+#              the recorded design check PASSes (default "false")
+#   idea_approved — "true" lets profile=auto accept the idea-eval report's recommendation
+#              (proceed/reshape → continue, drop → cancelled; missing → pause) (default "false")
+#   CLI --auto on dev-idea / dev-plan sets profile=auto AND both pre-clear vars to true.
 #   spurBin  — PATH-independent spur invocation (overridden by CLI at run start)
 #   agent    — agent for agent.run steps (default: auto → `agent.default` in config)
 #
@@ -137,15 +137,24 @@ states:
           session: fresh
           expectFile: .spur/run/${vars.__runId}-idea-eval-report.md
           timeoutMs: ${vars.stepTimeoutMs}
+      # Derive the eval recommendation ONCE (0769 pattern): the first non-blank line under
+      # `## Recommendation` → proceed | reshape | drop | unknown. The --auto guards out of
+      # discovery / idea-eval read this file; `unknown` routes to the idea-eval pause.
+      - kind: shell
+        options:
+          command: >-
+            awk '/^## Recommendation/{f=1;next} f&&NF{r=tolower($0);exit}
+            END{print (match(r,/^[*_ ]*(proceed|reshape|drop)/) ? substr(r,RSTART,RLENGTH) : "unknown")}'
+            ".spur/run/$__runId-idea-eval-report.md" | tr -d '*_ ' > ".spur/run/$__runId-idea-recommendation.txt"
 
   - id: idea-eval
     description: >
       HITL taste gate. Discovery has produced the idea-evaluation report
       (.spur/run/${vars.__runId}-idea-eval-report.md) with urgency/necessity scores, premises,
       pros/cons, and a recommendation. The operator reviews and approves or
-      rejects. This gate is NOT auto-clicked by --auto (taste gate, like
-      design-approval). Only routes around when vars.idea_approved=true
-      (explicit prior approval).
+      rejects. Routed around only when profile=auto AND vars.idea_approved=true
+      (CLI --auto) AND the report's recommendation is proceed/reshape; a drop
+      recommendation cancels, a missing one still pauses here.
     pause: true
     onEnter:
       - kind: hitl.confirm
@@ -181,16 +190,15 @@ states:
         options:
           path: .spur/run/${vars.__runId}-idea-feature-id.txt
           var: featureId
+      # 1007 R2: one shell persists both sections — the Goal and Scope updates share the
+      # artifact precondition, so the two update shells merged (original commands in order,
+      # same && fail-fast semantics).
       - kind: shell
         options:
           command: >-
             mkdir -p .spur/run &&
             test -s .spur/run/$__runId-idea-goal.md &&
-            $spurBin feature update "$featureId" --section Goal --from-file .spur/run/$__runId-idea-goal.md
-      - kind: shell
-        options:
-          command: >-
-            mkdir -p .spur/run &&
+            $spurBin feature update "$featureId" --section Goal --from-file .spur/run/$__runId-idea-goal.md &&
             test -s .spur/run/$__runId-idea-scope.md &&
             $spurBin feature update "$featureId" --section Scope --from-file .spur/run/$__runId-idea-scope.md
 
@@ -206,9 +214,10 @@ states:
       check records FAIL and routes through the capped retry loop via guards that consume the
       recorded result — it never fails the run (the engine's default onError policy is `fail`,
       so an in-action check failure would kill the run before the retry edges are evaluated).
-      Requirement coverage is MEASURED alongside it by the soft idea-coverage-check shell
-      (0887 R4): the recorded coverage status conjuncts into the profile=auto ac-generate
-      guards below so uncovered inventory items route through the same capped retry loop.
+      Requirement coverage is MEASURED by the same gate (1004 R2): the `idea-ac-check` args
+      carry `--inventory .spur/run/${vars.__runId}-idea-eval-report.md`, so uncovered
+      inventory items record FAIL on `idea-ac-check.status` and route through the same capped
+      retry loop via the guards below.
     onEnter:
       - kind: shell
         options:
@@ -246,33 +255,16 @@ states:
         options:
           id: idea-ac-check
           executable: "${vars.spurBin}"
-          args: ["feature", "check", "${vars.featureId}"]
+          args: ["feature", "check", "${vars.featureId}", "--inventory", ".spur/run/${vars.__runId}-idea-eval-report.md"]
           resultFile: .spur/run/${vars.__runId}-idea-ac-check.status
           softFail: true
           timeoutMs: 120000
-      # 0887 R4: requirement-inventory ↔ AC coverage, measured once at the same author/revise
-      # boundary. Soft shell (exit 0 always): the checker writes the PASS/FAIL status itself and
-      # the guards below consume the recorded result — never re-run the checker (0769 pattern).
-      # Repo-checkout path first, then the superskill-staged twin (handoff-finalize resolution
-      # shape); neither present fails closed to FAIL so readiness degrades visibly instead of
-      # silently skipping coverage.
-      - kind: shell
-        options:
-          command: >-
-            mkdir -p .spur/run &&
-            S=plugins/sp/scripts/idea-coverage-check.ts;
-            [ -f config/plugin-scripts.json -a -f "$S" ] || S="$(superskill script path sp idea-coverage-check.ts 2>/dev/null)";
-            if [ -n "$S" ] && [ -f "$S" ]; then
-              bun "$S" --run-id "$__runId" --report ".spur/run/$__runId-idea-eval-report.md" --ac ".spur/run/$__runId-idea-ac-content.md" || printf 'FAIL run=%s checker exited nonzero (bun missing or checker crash)\n' "$__runId" > ".spur/run/$__runId-idea-coverage.reason";
-            else
-              printf 'FAIL run=%s checker not found — run superskill install sp --marketplace gobing-ai/spur\n' "$__runId" | tee ".spur/run/$__runId-idea-coverage.reason" >&2;
-              printf 'FAIL\n' > ".spur/run/$__runId-idea-coverage.status";
-            fi
       # 0945 R2: derive the route facts ONCE at this boundary (0769 pattern) — guards read these
       # named files instead of re-deriving design×needs_design inline. The same command is
       # duplicated in feature-check's onEnter (YAML has no include); byte-equality is
       # test-pinned (idea-pipeline-routing.test.ts). Missing/corrupt needs-design JSON fails
-      # safe to `design`; readiness is PASS only when both recorded checks are PASS.
+      # safe to `design`; readiness is PASS only when the recorded check is PASS (1004 R2:
+      # coverage folds into idea-ac-check via --inventory).
       # (warn) 7 commands: two three-way `test && printf || printf` derivations over captured
       # signal files (ADR-115); no model hop, no CLI re-run.
       - kind: shell
@@ -282,7 +274,7 @@ states:
             { test "$design" = skip -o "$design" = auto -a "$(jq -r .needs_design .spur/run/$__runId-idea-needs-design.json 2>/dev/null)" = false &&
             printf 'skip\n' > .spur/run/$__runId-idea-design-route.txt ||
             printf 'design\n' > .spur/run/$__runId-idea-design-route.txt; } &&
-            { test "$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" = PASS -a "$(cat .spur/run/$__runId-idea-coverage.status 2>/dev/null)" = PASS &&
+            { test "$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" = PASS &&
             printf 'PASS\n' > .spur/run/$__runId-idea-ac-ready.status ||
             printf 'FAIL\n' > .spur/run/$__runId-idea-ac-ready.status; }
 
@@ -299,29 +291,19 @@ states:
       the transition guards route directly from ac-generate to the appropriate next
       state, so this state is only entered in interactive mode. On failure, the retry
       cap routes back to ac-generate (≤3 retries) or escalates to failed.
-      Requirement coverage (.spur/run/${vars.__runId}-idea-coverage.status, 0887 R4) is part of
-      the recorded results this gate surfaces: a FAIL there means some inventory items have no
-      covering scenario — answer no to route back to ac-generate for revision if needed.
+      Requirement coverage (1004 R2) folds into the recorded `idea-ac-check` result via the
+      gate's `--inventory` flag: a FAIL there can mean some inventory items have no covering
+      scenario — answer no to route back to ac-generate for revision if needed.
     pause: true
     onEnter:
       - kind: hitl.confirm
         options:
-          prompt: "Feature check for ${vars.featureId}. Review the AC and confirm to proceed? Requirement coverage status: $(cat .spur/run/${vars.__runId}-idea-coverage.status 2>/dev/null || echo unknown) — reason: $(cat .spur/run/${vars.__runId}-idea-coverage.reason 2>/dev/null || echo n/a) (.spur/run/${vars.__runId}-idea-coverage.status[.reason]). (Failures route back to ac-generate for revision, capped at 3 retries.)"
+          prompt: "Feature check for ${vars.featureId}. Review the AC and confirm to proceed? Recorded check status: $(cat .spur/run/${vars.__runId}-idea-ac-check.status 2>/dev/null || echo unknown) (.spur/run/${vars.__runId}-idea-ac-check.status — includes inventory-coverage findings, 1004 R2). (Failures route back to ac-generate for revision, capped at 3 retries.)"
           decision:
             mode: never
-      # 0945 R2: same writer as ac-generate's onEnter (byte-equal, test-pinned) — refreshes the
-      # derived route facts after the interactive gate so both entries into this boundary read
-      # consistent files; hitl.confirm only sets $__hitlAnswer.
-      - kind: shell
-        options:
-          command: >-
-            mkdir -p .spur/run &&
-            { test "$design" = skip -o "$design" = auto -a "$(jq -r .needs_design .spur/run/$__runId-idea-needs-design.json 2>/dev/null)" = false &&
-            printf 'skip\n' > .spur/run/$__runId-idea-design-route.txt ||
-            printf 'design\n' > .spur/run/$__runId-idea-design-route.txt; } &&
-            { test "$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" = PASS -a "$(cat .spur/run/$__runId-idea-coverage.status 2>/dev/null)" = PASS &&
-            printf 'PASS\n' > .spur/run/$__runId-idea-ac-ready.status ||
-            printf 'FAIL\n' > .spur/run/$__runId-idea-ac-ready.status; }
+      # 1007 R2: the former post-confirm route-writer duplicate is deleted — ac-generate's
+      # writer is the single writer (0945 R2 pin updated); its inputs (design var, needs-design
+      # JSON, recorded ac-check status) cannot change across hitl.confirm.
 
   - id: system-design
     description: >
@@ -376,9 +358,8 @@ states:
   - id: design-approval
     description: >
       HITL taste gate. The operator reviews the system design and approves or
-      rejects it. This gate is NOT auto-clicked by --auto (Auto-Decision Principle
-      #5: taste-decision -> surface to human). Only routes around when
-      vars.design_approved=true (explicit prior approval).
+      rejects it. Routed around only when profile=auto AND vars.design_approved=true
+      (CLI --auto) AND the recorded design check PASSes; a FAIL still pauses here.
 
       R7 (0433): a headless `no` persisted by DefaultHitlResponder would loop
       design-approval -> system-design -> design-approval, burning unbounded
@@ -559,10 +540,8 @@ states:
       - kind: shell
         options:
           command: >-
-            if [ -f packages/app/src/workflow/idea-handoff-cli.ts ]; then
-              bun packages/app/src/workflow/idea-handoff-cli.ts;
-            elif H="$(superskill script path sp idea-handoff.mjs 2>/dev/null)" && [ -f "$H" ]; then
-              node "$H";
+            eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/idea-handoff.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/idea-handoff.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; if [ -f "$S" ]; then
+              "$RUNNER" "$S";
             else
               echo "idea handoff failed closed — idea-handoff script not found — run 'superskill install sp --marketplace gobing-ai/spur'" >&2;
               exit 1;
@@ -575,9 +554,8 @@ states:
       next command (ready-depth refineall when any task is unready, else auto
       runall). No task execution — the pipeline stops here.
     onEnter:
-      - kind: note
-        options:
-          message: "Idea pipeline handoff. Feature: ${vars.featureId}. See .spur/run/${vars.__runId}-idea-handoff.md for the created task list and recommended next command."
+      # 1007 R2: the handoff note is deleted — the run.artifact below carries the handoff
+      # (same path, the recommended-next-command report).
       - kind: run.artifact
         options:
           path: .spur/run/${vars.__runId}-idea-handoff.md
@@ -608,17 +586,32 @@ transitions:
     guard:
       kind: always
 
-  # ── discovery: auto-skip (profile=auto + idea_approved) OR enter idea-eval taste gate ──
-  # Declaration order: auto-skip FIRST (same pattern as system-design → design-approval).
-  # Without this edge, discovery always enters idea-eval (pause: true), rendering the prompt
+  # ── discovery: auto-accept the eval recommendation (profile=auto + idea_approved) OR enter idea-eval ──
+  # Declaration order: auto edges FIRST (same pattern as system-design → design-approval).
+  # Without them, discovery always enters idea-eval (pause: true), rendering the prompt
   # and persisting an avoidable pause before the in-state auto-skip guard can fire (0366 R4/R5).
+  # --auto accepts the recommendation derived at the end of discovery (-idea-recommendation.txt):
+  # drop → cancelled, proceed/reshape → feature-create; unknown/missing falls through to the
+  # idea-eval pause (fail-safe: never auto-create a feature without an explicit proceed).
+  - from: discovery
+    to: cancelled
+    terminalReason: cancelled
+    description: profile=auto AND idea_approved=true AND eval recommends drop — accept it, no feature.
+    guard:
+      kind: shell
+      options:
+        command: >-
+          test "$profile" = auto && test "$idea_approved" = true &&
+          grep -qx drop .spur/run/$__runId-idea-recommendation.txt
   - from: discovery
     to: feature-create
-    description: profile=auto AND idea_approved=true — skip idea-eval taste gate entirely.
+    description: profile=auto AND idea_approved=true AND eval recommends proceed/reshape — skip idea-eval taste gate.
     guard:
       kind: shell
       options:
-        command: 'test "$profile" = auto && test "$idea_approved" = true'
+        command: >-
+          test "$profile" = auto && test "$idea_approved" = true &&
+          grep -qxE 'proceed|reshape' .spur/run/$__runId-idea-recommendation.txt
   - from: discovery
     to: idea-eval
     description: Discovery complete — gate on idea evaluation taste decision.
@@ -629,11 +622,13 @@ transitions:
   # Declaration order: auto-skip FIRST (same pattern as system-design → design-approval).
   - from: idea-eval
     to: feature-create
-    description: profile=auto AND idea_approved=true — skip idea-eval taste gate.
+    description: profile=auto AND idea_approved=true AND eval recommends proceed/reshape — skip idea-eval taste gate.
     guard:
       kind: shell
       options:
-        command: 'test "$profile" = auto && test "$idea_approved" = true'
+        command: >-
+          test "$profile" = auto && test "$idea_approved" = true &&
+          grep -qxE 'proceed|reshape' .spur/run/$__runId-idea-recommendation.txt
   - from: idea-eval
     to: feature-create
     description: Idea evaluation approved — create or select a feature.
@@ -694,32 +689,32 @@ transitions:
           test "$profile" = auto &&
           test "$(cat .spur/run/$__runId-idea-ac-ready.status 2>/dev/null)" = PASS &&
           test "$(cat .spur/run/$__runId-idea-design-route.txt 2>/dev/null)" = skip
-  # Auto-skip 3: check or coverage failed, retry < 3 → loop back to ac-generate (self-loop)
+  # Auto-skip 3: check failed (incl. inventory coverage), retry < 3 → loop back to ac-generate (self-loop)
   - from: ac-generate
     to: ac-generate
-    description: "profile=auto, check or coverage failed, retry cap not reached — re-run ac-generate."
-    # (warn) 5 commands: one persistent multi-assignment line captures both statuses and the retry count
-    # (with its 0 fallback); the failed-check pair stays verbatim and the profile gate folds into the retry
-    # test via `test -a`; the retry loop is the smallest honest formulation of cap<3 routing (0769).
+    description: "profile=auto, check failed (incl. inventory coverage), retry cap not reached — re-run ac-generate."
+    # (warn) 4 commands: one persistent multi-assignment line captures the check status and the retry
+    # count (with its 0 fallback); the profile gate folds into the retry test via `test -a`; the retry
+    # loop is the smallest honest formulation of cap<3 routing (0769).
     guard:
       kind: shell
       options:
         command: >-
-          ac_status="$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" cov_status="$(cat .spur/run/$__runId-idea-coverage.status 2>/dev/null)" retry="$(cat .spur/run/$__runId-idea-ac-retry-count 2>/dev/null || echo 0)";
-          { test "$ac_status" != PASS || test "$cov_status" != PASS; } && test "$profile" = auto -a "$retry" -lt 3
-  # Auto-skip 4: check or coverage failed, retry cap reached → escalate to failed
+          ac_status="$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" retry="$(cat .spur/run/$__runId-idea-ac-retry-count 2>/dev/null || echo 0)";
+          test "$ac_status" != PASS && test "$profile" = auto -a "$retry" -lt 3
+  # Auto-skip 4: check failed (incl. inventory coverage), retry cap reached → escalate to failed
   - from: ac-generate
     to: failed
     terminalReason: retry-exhausted
-    description: "profile=auto, check or coverage failed after 3 retries — escalate to failed."
-    # (warn) 5 commands: mirror of the retry guard with cap>=3; keeping escalation and retry as one
+    description: "profile=auto, check failed (incl. inventory coverage) after 3 retries — escalate to failed."
+    # (warn) 4 commands: mirror of the retry guard with cap>=3; keeping escalation and retry as one
     # test-chain pair makes the cap boundary auditable in the diff (0769).
     guard:
       kind: shell
       options:
         command: >-
-          ac_status="$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" cov_status="$(cat .spur/run/$__runId-idea-coverage.status 2>/dev/null)" retry="$(cat .spur/run/$__runId-idea-ac-retry-count 2>/dev/null || echo 0)";
-          { test "$ac_status" != PASS || test "$cov_status" != PASS; } && test "$profile" = auto -a "$retry" -ge 3
+          ac_status="$(cat .spur/run/$__runId-idea-ac-check.status 2>/dev/null)" retry="$(cat .spur/run/$__runId-idea-ac-retry-count 2>/dev/null || echo 0)";
+          test "$ac_status" != PASS && test "$profile" = auto -a "$retry" -ge 3
   # Interactive fallback: enter feature-check HITL gate
   - from: ac-generate
     to: feature-check

package/config/workflows/pr-review.yaml

@@ -50,6 +50,7 @@ vars:
   noWait: "false"
   waitTimeoutSec: "600"
   waitIntervalSec: "30"
+  spurBin: "spur"
   preReviewCmd: ""
   __runId: ""
   # Runtime-set by file.read.into-var (0771): the request record's requestedAt|head,
@@ -65,15 +66,23 @@ states:
       before any push can publish it).
       Writes PASS|FAIL to .spur/run/${vars.__runId}-pr-preflight.status; always exit 0.
     onEnter:
+      # 1007 R8: run-start script-resolution identity (same command as feature-verification
+      # verify's first action) — records which tree this run resolved (source-repo vs
+      # installed) into .spur/run/$__runId-script-root.json; later plugin-script actions
+      # resolve through it (1007 R4 snippet). Never aborts.
+      - kind: shell
+        options:
+          command: >-
+            S=plugins/sp/scripts/script-root.ts; [ -f config/plugin-scripts.json -a -f "$S" ] || S="$(superskill script path sp script-root.mjs 2>/dev/null)"; RUNNER=bun; case "$S" in *.mjs) RUNNER=node ;; esac; [ -f "$S" ] && "$RUNNER" "$S" --run-id "$__runId" --spur-bin "$spurBin" || echo "script-root failed closed — run 'superskill install sp --marketplace gobing-ai/spur'" >&2; exit 0
       - kind: shell
         options:
           # (e) 0825: 10 lines — detached-HEAD/dirty/gh-auth/base-refusal probes must record
           # before any push can publish; split would let a half-probed run continue.
           command: >-
-            mkdir -p .spur/run &&
+            
             STATUS_FILE=".spur/run/$__runId-pr-preflight.status" &&
             set +e &&
-            bun "$(superskill script path sp pr-reviewing.ts)" preflight --base "$baseBranch" --json > ".spur/run/$__runId-pr-context.json";
+            eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" preflight --base "$baseBranch" --json > ".spur/run/$__runId-pr-context.json";
             rc=$?; set -e &&
             if [ "$rc" -eq 0 ]; then printf 'PASS\n' > "$STATUS_FILE"; else printf 'FAIL\n' > "$STATUS_FILE"; fi &&
             exit 0
@@ -88,10 +97,10 @@ states:
           # (e) 0825: 9 lines — hygiene verdict + json capture land together; the status-file
           # fallback keeps a silent driver from passing the gate.
           command: >-
-            mkdir -p .spur/run &&
+            
             STATUS_FILE=".spur/run/$__runId-pr-hygiene.status" &&
             set +e &&
-            bun "$(superskill script path sp pr-reviewing.ts)" hygiene --base "$baseBranch" --json
+            eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" hygiene --base "$baseBranch" --json
             --status-file "$STATUS_FILE" > ".spur/run/$__runId-pr-hygiene.json";
             rc=$?; set -e &&
             if [ ! -f "$STATUS_FILE" ]; then printf 'FAIL\n' > "$STATUS_FILE"; fi &&
@@ -122,7 +131,7 @@ states:
             mkdir -p .spur/run &&
             STATUS_FILE=".spur/run/$__runId-pr-push.status" &&
             set +e &&
-            bun "$(superskill script path sp pr-reviewing.ts)" push --json --status-file "$STATUS_FILE"
+            eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" push --json --status-file "$STATUS_FILE"
             > ".spur/run/$__runId-pr-push.json";
             rc=$?; set -e &&
             if [ ! -f "$STATUS_FILE" ]; then printf 'FAIL\n' > "$STATUS_FILE"; fi &&
@@ -139,7 +148,7 @@ states:
             mkdir -p .spur/run &&
             STATUS_FILE=".spur/run/$__runId-pr-ensure.status" &&
             set +e &&
-            bun "$(superskill script path sp pr-reviewing.ts)" ensure-pr --base "$baseBranch" --json
+            eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" ensure-pr --base "$baseBranch" --json
             --status-file "$STATUS_FILE" > ".spur/run/$__runId-pr.json";
             rc=$?; set -e &&
             if [ ! -f "$STATUS_FILE" ]; then printf 'FAIL\n' > "$STATUS_FILE"; fi &&
@@ -155,7 +164,7 @@ states:
           # (e) 0825: 10 lines — the write-then-read split is the file-var transport feeding
           # wait/collect and must stay one action (same record, same run scope).
           command: >-
-            mkdir -p .spur/run; case "$mode" in rerun) F=--force ;; *) F= ;; esac; bun "$(superskill script path sp pr-reviewing.ts)" request --focus "$focus" $F --json --status-file ".spur/run/$__runId-pr-request.status" > ".spur/run/$__runId-pr-request.json"; [ -f ".spur/run/$__runId-pr-request.status" ] || printf 'FAIL\n' > ".spur/run/$__runId-pr-request.status"; jq -e . ".spur/run/$__runId-pr-request.json" > /dev/null 2>&1 && jq -r '.requestedAt // ""' ".spur/run/$__runId-pr-request.json" > ".spur/run/$__runId-pr-since.txt" && jq -r '.head // ""' ".spur/run/$__runId-pr-request.json" > ".spur/run/$__runId-pr-head.txt"; exit 0
+            F="$([ "$mode" = rerun ] && echo --force)"; eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" request --focus "$focus" $F --json --status-file ".spur/run/$__runId-pr-request.status" > ".spur/run/$__runId-pr-request.json"; [ -f ".spur/run/$__runId-pr-request.status" ] || echo FAIL > ".spur/run/$__runId-pr-request.status"; jq -e . ".spur/run/$__runId-pr-request.json" >/dev/null 2>&1 && jq -r '.requestedAt // ""' ".spur/run/$__runId-pr-request.json" > ".spur/run/$__runId-pr-since.txt" && jq -r '.head // ""' ".spur/run/$__runId-pr-request.json" > ".spur/run/$__runId-pr-head.txt"; exit 0
 
   - id: wait
     description: >
@@ -180,7 +189,7 @@ states:
             mkdir -p .spur/run &&
             STATUS_FILE=".spur/run/$__runId-pr-wait.status" &&
             set +e &&
-            bun "$(superskill script path sp pr-reviewing.ts)" wait --since "$prSince" --head "$prHead" --timeout "$waitTimeoutSec" --interval "$waitIntervalSec"
+            eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" wait --since "$prSince" --head "$prHead" --timeout "$waitTimeoutSec" --interval "$waitIntervalSec"
             --json --status-file "$STATUS_FILE" > ".spur/run/$__runId-pr-wait.json";
             rc=$?; set -e &&
             if [ ! -f "$STATUS_FILE" ]; then printf 'FAIL\n' > "$STATUS_FILE"; fi &&
@@ -204,7 +213,7 @@ states:
           # (e) 0825: 9 lines — collect + status must run against the same since/head pair
           # before the composite status is decided (same 0771 record contract).
           command: >-
-            mkdir -p .spur/run; bun "$(superskill script path sp pr-reviewing.ts)" collect --since "$prSince" --head "$prHead" --json --status-file ".spur/run/$__runId-pr-collect.status" > ".spur/run/$__runId-pr-findings.json"; c=$?; bun "$(superskill script path sp pr-reviewing.ts)" status --since "$prSince" --head "$prHead" --json > ".spur/run/$__runId-pr-status.json" && [ "$c" -eq 0 ] && [ -f ".spur/run/$__runId-pr-collect.status" ] || printf 'FAIL\n' > ".spur/run/$__runId-pr-collect.status"; cat ".spur/run/$__runId-pr-findings.json" 2>/dev/null; exit 0
+            mkdir -p .spur/run; eval "$(jq -r 'if .mode=="source-repo" then "S="+@sh"\(.dir)/pr-reviewing.ts"+" RUNNER=bun" elif .mode=="installed" then "S="+@sh"\(.dir)/pr-reviewing.mjs"+" RUNNER=node" else empty end' ".spur/run/$__runId-script-root.json" 2>/dev/null)"; "$RUNNER" "$S" collect --since "$prSince" --head "$prHead" --json --status-file ".spur/run/$__runId-pr-collect.status" > ".spur/run/$__runId-pr-findings.json"; c=$?; "$RUNNER" "$S" status --since "$prSince" --head "$prHead" --json > ".spur/run/$__runId-pr-status.json" && [ "$c" -eq 0 ] && [ -f ".spur/run/$__runId-pr-collect.status" ] || printf 'FAIL\n' > ".spur/run/$__runId-pr-collect.status"; cat ".spur/run/$__runId-pr-findings.json" 2>/dev/null; exit 0
 
   - id: done
     description: Terminal — review collected; findings artifact at .spur/run/${vars.__runId}-pr-findings.json