fix(gate): unbreak all three of main's red gates in one PR, because one is not enough

AGENTS.md

@@ -73,12 +73,17 @@ make gate                # = no-scratch + no-secrets + design-refs
                          #   + self-driving-test (the shell harness)
 ```
 
-That is twenty-four steps, and the list is not maintained by hand: it is
-`GATE_STEPS` in the `Makefile`, and `gate-parity` (`scripts/check-gate-parity.sh`)
-fails if this block or CONTRIBUTING.md's stops matching it. The block had
-already drifted twice before that guard existed, both times by under-reporting
-a newly added guard, which is the direction that misleads — a reader runs the
-short list, sees green, and believes the gate is green (#1437).
+That list is not maintained by hand: it is `GATE_STEPS` in the `Makefile`, and
+`gate-parity` (`scripts/check-gate-parity.sh`) fails if this block or
+CONTRIBUTING.md's stops matching it. The block had already drifted twice before
+that guard existed, both times by under-reporting a newly added guard, which is
+the direction that misleads — a reader runs the short list, sees green, and
+believes the gate is green (#1437).
+
+Deliberately no total: the guard checks each step by name, so two PRs adding
+different guards merge cleanly, while a spelled-out count is one shared cell
+both must write — and two of them collided on it twice in a day, each time
+leaving `main` red for everyone (#1883).
 
 CI enforces the same steps split across three workflows:
 `/.github/workflows/ci.yml`'s required job runs everything except `invariants`

CONTRIBUTING.md

@@ -95,7 +95,7 @@ cargo build -p stella-cli --bin stella && \
   STELLA_BIN="$PWD/target/debug/stella" ./scripts/test-self-driving.sh
 ```
 
-Or just `make gate`, which is the twenty-four of them in order.
+Or just `make gate`, which runs all of them in order.
 
 Do not maintain that list by hand. It is `GATE_STEPS` in the `Makefile`, and
 `./scripts/check-gate-parity.sh` — itself one of the steps — fails if this fence

crates/stella-pipeline/src/pipeline/witness_stage.rs

@@ -52,7 +52,7 @@ fn apply_role_shaping(mut config: EngineConfig, overrides: &RoleCallOverrides) -
         config.max_output_tokens = Some(max_output_tokens);
     }
     if let Some(params) = &overrides.params {
-        config.params = Some(params.clone());
+        config.params = Some(*params);
     }
     config
 }
@@ -645,10 +645,12 @@ mod tests {
     /// `EngineConfig` knob — the type system keeps it raw-call-scoped).
     #[test]
     fn verifier_shaping_overlays_the_worker_engine_config() {
-        let mut worker = EngineConfig::default();
-        worker.temperature = Some(0.7);
-        worker.max_output_tokens = Some(1000);
-        worker.effort = None;
+        let worker = EngineConfig {
+            temperature: Some(0.7),
+            max_output_tokens: Some(1000),
+            effort: None,
+            ..EngineConfig::default()
+        };
 
         let shaped = apply_role_shaping(
             worker.clone(),

scripts/check-gate-parity.sh

@@ -66,26 +66,25 @@ if [ "$count" -eq 0 ]; then
   exit 1
 fi
 
-# Spelled-out numbers, because both documents spell the count in prose ("the
-# fifteen of them in order") and a digit would read wrong there. Covers a range
-# no plausible gate will leave.
-number_word() {
-  case "$1" in
-  10) echo ten ;; 11) echo eleven ;; 12) echo twelve ;; 13) echo thirteen ;;
-  14) echo fourteen ;; 15) echo fifteen ;; 16) echo sixteen ;;
-  17) echo seventeen ;; 18) echo eighteen ;; 19) echo nineteen ;;
-  20) echo twenty ;; 21) echo twenty-one ;; 22) echo twenty-two ;;
-  23) echo twenty-three ;; 24) echo twenty-four ;; 25) echo twenty-five ;;
-  *) echo "" ;;
-  esac
-}
-
-word="$(number_word "$count")"
-if [ -z "$word" ]; then
-  note "FAIL — $count gate steps is outside the range number_word() spells."
-  note "     Extend the table in this script; the documents spell the count."
-  exit 1
-fi
+# ── The count is deliberately NOT checked any more ───────────────────────────
+#
+# Both documents used to spell the total in prose ("the fifteen of them in
+# order") and this guard held them to it. That check is gone, and its removal
+# is the fix for a failure it caused rather than caught (#1883).
+#
+# The step NAMES are checked one at a time, so two PRs that each add a
+# different guard produce diffs that merge cleanly. The TOTAL is a single
+# shared cell both branches must write — and each writes its own correct
+# answer. On 2026-08-06 `module-reachability` and `self-driving-test` landed
+# within an hour of each other, each having dutifully updated both documents
+# to "twenty-four". The second merge left GATE_STEPS at 25 with the prose
+# saying 24, `docs guards` went red on `main`, and every open PR inherited it.
+# Twice in one day.
+#
+# Nothing was lost by dropping it. The count told a reader no fact the checked
+# list does not already carry: if every step is named, the number of them is
+# not independently knowable-wrong. It was a derived value maintained by hand,
+# which is the same defect this guard exists to prevent one level up.
 
 # CONTRIBUTING.md lists raw commands rather than make targets. Every guard is
 # `scripts/check-<target>.sh` except the ones named here, and the four compile
@@ -130,11 +129,6 @@ for doc in "$agents" "$contributing"; do
     fi
   done
 
-  if ! grep -qF -- "$word" "$doc"; then
-    note "FAIL — $doc does not spell the gate's step count as '$word'."
-    note "     The gate runs $count steps. Find the stale count and fix it."
-    fail=1
-  fi
 done
 
 # ── Ghosts ───────────────────────────────────────────────────────────────────
@@ -176,5 +170,5 @@ if [ "$fail" -ne 0 ]; then
   exit 1
 fi
 
-printf 'check-gate-parity: OK — %s (%s) gate steps, named in %s and %s.\n' \
-  "$count" "$word" "$agents" "$contributing"
+printf 'check-gate-parity: OK — all %s gate steps are named in %s and %s.\n' \
+  "$count" "$agents" "$contributing"

scripts/file-size-baseline.txt

@@ -17,25 +17,25 @@
 1659 bench/terminal_bench_analysis/tests/test_tb21_evidence_contract.py
 2269 crates/stella-cli/src/agent.rs
 1751 crates/stella-cli/src/agent/tests.rs
-4740 crates/stella-cli/src/command_deck.rs
-1507 crates/stella-cli/src/fleet_cmd.rs
+4691 crates/stella-cli/src/command_deck.rs
+1504 crates/stella-cli/src/fleet_cmd.rs
 2129 crates/stella-core/src/bus.rs
 2580 crates/stella-core/src/driver.rs
 3680 crates/stella-core/src/driver/tests.rs
 1780 crates/stella-model/src/anthropic/tests.rs
 2093 crates/stella-model/src/openai.rs
 1565 crates/stella-model/src/zai.rs
 1895 crates/stella-model/src/zai/tests.rs
-3642 crates/stella-pipeline/src/pipeline.rs
+3580 crates/stella-pipeline/src/pipeline.rs
 2568 crates/stella-pipeline/src/pipeline/tests.rs
-2962 crates/stella-protocol/src/event.rs
+2964 crates/stella-protocol/src/event.rs
 1995 crates/stella-store/src/lib.rs
 2268 crates/stella-store/src/tests.rs
 1916 crates/stella-store/src/usage.rs
 1569 crates/stella-tools/src/media.rs
 2184 crates/stella-tools/src/registry.rs
 1839 crates/stella-tools/src/scripts.rs
-1528 crates/stella-tui/src/deck_render.rs
+1531 crates/stella-tui/src/deck_render.rs
 4006 crates/stella-tui/src/deck_ui.rs
 1665 crates/stella-tui/src/views/engine.rs
 1611 crates/stella-tui/src/views/session.rs