feat(memory): missions remember their verdicts, per repository

Until now missions wrote no memory. The chat path records every turn into
the claw's .brain, but a mission's crew is minted per mission, so a brain
keyed by agent would be written once and never read. What persists across
missions is the repository: mission_memory keeps one .brain per repo_id,
writes each judge verdict into it (reason when met, sanitized guidance when
not — the operator reason may quote the acceptance text), and recalls
against the next phase's task text into the brief, under a heading all
three executors carry because it rides on the task.

Recall is BM25 over the keyword index, no embedder; the harness asserts the
brief carries the section once the repo has one judged mission behind it,
and says 'first mission' rather than failing before that. OpenClaw's
flush-before-compaction was the other half of this item and is moot here:
the chat loop has no compaction and already remembers both halves of
every turn.

Co-Authored-By: Claude Opus 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01WZb5A2kfVfjpdwSochkuHz

scripts/verify-mission-delivery.sh

@@ -355,6 +355,9 @@ assert_microvm() { # <token> <mission> <report>
   # calls at all, and none of them may have been a write.
   assert_verifier_read_only "$mission" microvm
 
+  # Project memory: the brief carries earlier verdicts on this repository.
+  assert_project_memory "$mission" microvm
+
   # And the verdict: judged, and by whom. `independent` is only true when the
   # judge came from a different provider family than the agent.
   indep=$(ssh "$HOST" "docker exec clawmates_postgres_1 psql -U postgres -d clawmates -tAc \
@@ -900,6 +903,34 @@ assert_verifier_read_only() { # <mission> <label>
   esac
 }
 
+# Did this mission's brief carry what earlier missions on the same repository
+# learned? `mission_memory` writes every judge verdict into a per-repo brain
+# and recalls against the next phase's task, so once the repo has ONE judged
+# mission behind it, the section must be in the prompt. Two branches, because
+# the first mission on a repo has nothing to recall and must not fail for it:
+# the earlier-verdict count is what decides which branch is the truth.
+assert_project_memory() { # <mission> <label>
+  local counts earlier carried
+  counts=$(ssh "$HOST" "docker exec clawmates_postgres_1 psql -U postgres -d clawmates -tAc \
+    \"select (select count(*) from mission_phase_evaluations e
+                 join missions m on m.id = e.mission_id
+                where m.repo_id = (select repo_id from missions where id='$1')
+                  and e.mission_id <> '$1' and e.error is null
+                  and e.created_at < (select created_at from missions where id='$1'))::text
+            || ' ' ||
+            (select count(*) from mission_events
+              where mission_id='$1' and kind='prompt.composed'
+                and detail->>'text' like '%What past missions on this repository learned%')::text;\"" \
+    | head -1 | tr -d '\r')
+  earlier=${counts%% *}; carried=${counts##* }
+  case "$earlier:$carried" in
+    0:0)  pass "$2-memory: first judged mission on this repo — nothing to recall, nothing carried" ;;
+    0:*)  fail "$2-memory: the brief carried project memory but no earlier verdict exists for this repo" ;;
+    *:0)  fail "$2-memory: $earlier earlier verdict(s) on this repo and the brief carried NONE — recall is not reaching the prompt" ;;
+    *)    pass "$2-memory: $earlier earlier verdict(s) on this repo; the brief carried the recalled section" ;;
+  esac
+}
+
 # ── Scenario: a model sizes the team ─────────────────────────────
 #
 # Slice 5. The planner proposes a roster for THIS mission, a human approves it,