humanish 0.126.0 → 0.127.1

package/docs/contracts/schemas.md

@@ -247,16 +247,19 @@ The rest of this section names keys by their v3 spelling:
   against the sandbox ceiling. A rerun starts its selected participants together;
 - `participants[].endAfterMs` and `windDownMs` (computer-use only): when the participant's
   session ends, in whole milliseconds on the `startAfterMs` clock and after its start, and how
-  long before that each turn tells it the end time and the minutes left (default 600,000).
+  long before that each turn tells it the end time and the minutes left (default a quarter of
+  the window, at most 600,000; `participantSchedule` in `src/study/arrivals.ts`).
   Refused: an end at or before the start, a `windDownMs` without `endAfterMs`, an end on
   `shared-world`, and a window whose sandbox deadline passes the sandbox ceiling. The
-  participant's session budget is its window plus 5 minutes (`participantSessionMs` in
+  participant's session budget is its window plus 5 minutes (`participantSession` in
   `src/study/arrivals.ts`) in place of `execution.timeoutMs`, and its sandbox deadline adds the
   route's headroom to that. At the end the loop requests the closing report (`debrief.trigger:
   scheduled_end`) and the session ends `goal_satisfied` with `stopCause: scheduled_end`. The
-  outcome an accepted report declares, its `outcome` field or the first line of its summary, is
-  the trace's `declaredOutcome`. The blocker rule reads a declared `reached` or `blocked` before
-  the summary, and after a declared `not_reached` it reads the summary. Before
+  outcome an accepted report declares, its `outcome` field or the first line of its summary that
+  is not a `>` blockquote, is the trace's `declaredOutcome`. The blocker rule reads a declared
+  `reached` or `blocked` before the summary, and after a declared `not_reached` it reads the
+  summary. A participant that passes without declaring `reached` counts in the review's
+  `participants.ranUntilEnd`, and its completion labels say it ran until its scheduled end. Before
   the end, a turn that requests no action does not end the session: the trace records a
   `participant ended before its scheduled end` notice, the next request says the time left, and
   the turn counts toward the idle backstop. Before the end, a turn that only waits or takes
@@ -296,16 +299,21 @@ The rest of this section names keys by their v3 spelling:
   `HUMANISH_E2B_MAX_CONCURRENT_SANDBOXES` sandboxes at once (20 when unset; a provisioned
   shared world's app sandbox counts against it), so a larger roster runs in waves and the plan
   records a warning naming the setting. A declared value above that limit is refused at plan
-  time. Total sessions and spend are identical either way; only wall-clock
-  and simultaneity differ. Declaring a value below the participant count runs participants in
+  time, unless `HUMANISH_CUA_MAX_CONCURRENCY` lowers it first; the limit then lowers what is
+  left as it does an undeclared default. Total sessions and spend are identical either way; only
+  wall-clock and simultaneity differ. Declaring a value below the participant count runs participants in
   waves and emits a warning saying so, because a green waved run is otherwise
   indistinguishable from the all-live run the author meant. Shared-world studies
   need at least 2 (the host and a follower are live together); `concurrency: 1`
   there is refused at parse, since the sequential turn-taking route was removed
   in 0.106.0. The env override
   `HUMANISH_CUA_MAX_CONCURRENCY` may only lower the effective bound, never
-  raise concurrent paid desktops, and a lowering is recorded on
-  the plan (`envLoweredConcurrencyFrom`). Inert (warned) on other routes.
+  raise concurrent paid desktops. The plan applies it before the E2B limit, on every
+  computer-use desktop and on shared-world, so the schedule, `study check` and the late-start
+  warnings use the lowered count; the plan records a warning and, on computer-use,
+  `envLoweredConcurrencyFrom`. When the E2B limit lowers the count further, the plan records
+  the E2B limit's warning instead. A shared world refuses a value under 2. Inert
+  (warned) on other routes.
   `execution.timeoutMs` is the per-participant session budget on this route (semantics
   change: it was the single-session budget pre-fan-out); there is no run-level
   wall clock. `policies.allowPublicTargets` cannot combine with N>1 against one
@@ -1914,8 +1922,13 @@ verification or maintainer acceptance.
 the run passed also reads `run.json`'s `outcome.ok` ([Run Outcome](run-bundle.md#run-outcome)):
 a run whose execution failed shows as failed beside a `pass` verdict. `participants.reachedGoal` retains the count of
 recorded successful sessions; it is not an independent adjudication of their
-claims. Computer-use review and Observer labels distinguish a participant's
-reported completion from a recorded `stopWhen` or dwell condition match. A
+claims. A computer-use session that passed at its scheduled end or session cap
+counts there only when its participant declared `reached`; otherwise it counts
+in `participants.ranUntilEnd`, present when above 0. Bundles written before
+`ranUntilEnd` existed counted it in `reachedGoal`, and a re-read keeps that
+count. Computer-use review and Observer labels distinguish a participant's
+reported completion from a recorded `stopWhen` or dwell condition match, and
+from a pass at a clock end without a declared reach. A
 condition match establishes that condition, not every aspect of the mission.
 Missing or malformed completion evidence is labeled unavailable. Other actor
 routes retain their own completion semantics. Re-reading a historical review

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "humanish",
-  "version": "0.126.0",
+  "version": "0.127.1",
   "description": "Synthetic user research for apps, CLIs, and agent-facing product flows.",
   "keywords": [
     "cli",