toolroll 0.9.74 → 0.9.75

package/CHANGELOG.md

@@ -1,5 +1,36 @@
 # Changelog
 
+## 0.9.75 — 2026-10-27
+
+- **Opening a page only shows it.** Pages no longer run checks, tidy up or
+  call out as you open them. Sign-ins, integrations, update checks, storage
+  and Codex's tools are checked in the background, and Models, Updates,
+  Storage, Pull requests, Tools, Project setup and the GitHub listing show
+  their last answer and when it was checked, with a **Check now** button to
+  ask again. On a coding session, **Review for shipping** and the changes
+  view are buttons, so a link preview or prefetch can't run them. Reading a
+  checkout an agent wrote never runs that checkout's own git settings.
+- **Reading an agent's checkout can't be raced into running a filter.** Git
+  reads the checkout through a private copy of its settings taken once, so a
+  filter the agent adds to its git config and `.git/info/attributes` after
+  that copy never runs.
+- **The task card never contradicts itself.** One answer to "did the
+  approved check pass" across the task page, the road, the Tasks list, the
+  result page and chat. A check whose saved log can't be read is never
+  shown as passed, and a check whose command was approved again since it ran
+  says so: "The check command changed since this check; run checks again."
+- **Lead-filed runs nest under the lead on Agents → Live.** Runs your own
+  lead filed sit under its tile. Once its turn has gone, they stay grouped
+  under a quiet "Filed by lead chat" header. The lead's full screen lists
+  every task it filed. Only you see what your lead filed.
+- **The lead passes asks to a helper instead of answering for them.** "Ask
+  Maya to …" now reaches Maya even when she seems to lack a tool or the
+  knowledge for it; she says so herself or asks you first.
+- **A helper's "can't" reaches whoever asked.** When a helper can't do
+  something asked at its desk, its note (or its reason) goes back to the
+  person or agent who asked, instead of waiting in an inbox.
+- Database: schema v136 (unchanged).
+
 ## 0.9.74 — 2026-10-27
 
 - **Agents → Live: every agent and subagent at a glance.** A read-only grid

package/dist/assignment.d.ts

@@ -28,6 +28,9 @@ export type AssignmentChecks = {
     batch?: ChecksBatch | null;
     /** Checks Off: the passing full release check that contains this commit (release-coverage.ts), read now, never sealed. */
     release?: ReleaseCoverage | null;
+    /** Why a saved check can't be read, when it is the project's check command approved again since it ran (its files
+     * are intact; running checks again settles it). Presentation only, never sealed. */
+    cause?: "command-changed" | null;
 };
 export type AssignmentReceipt = {
     digest: string;

package/dist/assignment.js

@@ -21,7 +21,7 @@ import { requiredCheckCommandFor, runCheckLevel } from "./check-levels.js";
 import { followUpChecksOf, withFollowUps } from "./result-follow-ups.js";
 import { buildReviewOf, findingWords } from "./review-switch.js";
 import { NEEDS, WAITS, processNeedOf, resultHoldUpSentence } from "./needs-you.js";
-import { assignmentStageOf, checkBackingOf, completionBlockersOf, requirementWordOf } from "./task-status.js";
+import { assignmentStageOf, checkBackingOf, completionBlockersOf, receiptChecksOf, requirementWordOf } from "./task-status.js";
 import { releaseCoverageOf } from "./release-coverage.js";
 import { leadClaimOf } from "./lead-voice.js";
 /** Status-first handoff for routine reads. Fetch get_assignment only when
@@ -53,7 +53,7 @@ function olderStoreSafe(read, absent) {
     }
 }
 /** A check as the receipt digest seals it: presentation-only fields (check-levels.ts) left out. */
-const sealedChecks = ({ level: _level, running: _running, batch: _batch, release: _release, ...proof }) => proof;
+const sealedChecks = ({ level: _level, running: _running, batch: _batch, release: _release, cause: _cause, ...proof }) => proof;
 const OWNER_ACTION = "assignment claimed";
 const CHECK_ACTION = COMPLETION_ACTION;
 const actorOf = (owner) => `${owner.kind}:${owner.id}`;
@@ -87,7 +87,7 @@ function ownerOf(store, rootId, repo) {
 // Report the recorded check outcome, never turn a model verdict into a check.
 // Follow-up checks on the same commit (Run checks) upgrade a pass and show a failure.
 export function assignmentChecksForRun(store, root, runId, now) {
-    const own = buildChecks(store, root, runId);
+    const own = recordedFailureStands(store, runId, buildChecks(store, root, runId));
     const run = store.getRun(runId);
     const followUps = followUpChecksOf(store, runId, now, root);
     if (followUps.length === 0)
@@ -106,6 +106,18 @@ export function assignmentChecksForRun(store, root, runId, now) {
         ...(changed ? { status: read.status, level: read.level, exitCode: read.exitCode, logArtifactId: latest?.logArtifactId ?? own.logArtifactId,
             detail: read.status === "passed" ? `${words} passed.` : `${words} failed (exit ${read.exitCode}).` } : {}) };
 }
+/** A failure the builder recorded (run_check) stands when its saved receipt can't be read now or says nothing ran:
+ * fail closed, every surface reads the same known failure and its exit code (task-status.ts approvedCheckOf). */
+function recordedFailureStands(store, runId, own) {
+    if (own.status !== "unavailable" && own.status !== "not-run")
+        return own;
+    const recorded = olderStoreSafe(() => store.runCheckFor(runId), null);
+    if (recorded?.status !== "failed")
+        return own;
+    return { ...own, status: "failed", exitCode: recorded.exitCode,
+        detail: `${own.level === "quick" ? "Quick checks" : "Checks"} failed${recorded.exitCode === null ? "" : ` (exit ${recorded.exitCode})`}. ${own.detail}` };
+}
+const COMMAND_MISSING_OR_CHANGED = "The original approved verification command is missing or changed.";
 function buildChecks(store, root, runId) {
     const level = runCheckLevel(store, runId);
     const unavailable = (detail) => ({ status: "unavailable", exitCode: null, command: null, logArtifactId: null, detail, level });
@@ -113,8 +125,13 @@ function buildChecks(store, root, runId) {
         return unavailable("Saved checks are unavailable.");
     try {
         const gate = verificationEvidence(store, root, runId);
-        if (!gate.ok)
-            return unavailable(gate.problem);
+        if (!gate.ok) {
+            // The approved command changed since this check ran: nothing is missing, the check just no longer binds to it.
+            const run = store.getRun(runId), repo = run === null ? null : store.refById(run.taskRef)?.repo ?? null;
+            const live = repo === null ? null : store.liveVerifyCommand(repo);
+            const changed = gate.problem === COMMAND_MISSING_OR_CHANGED && live !== null && run !== null && live.approvedAt > run.startedAt;
+            return changed ? { ...unavailable(gate.problem), cause: "command-changed" } : unavailable(gate.problem);
+        }
         const receipt = gate.bytes === null ? null : JSON.parse(gate.bytes);
         if (receipt === null)
             return { ...unavailable(level === "off" ? "Checks were off for this build." : "No machine check is recorded."), status: "not-run" };
@@ -390,10 +407,12 @@ function readinessOf(store, receipt, repo, review) {
     const high = review === null || review.sentBackAs !== null || review.state !== "reviewed" ? 0 : review.high.length;
     const checkRequired = olderStoreSafe(() => requiredCheckCommandFor(store, repo, receipt.taskId, receipt.checks.level), null) !== null;
     const matrix = receipt.proof?.matrix ?? [];
-    const backing = checkBackingOf(receipt.checks);
-    const verdict = shownVerdictOf(receipt.proof?.verdict ?? null, { checkFailed: !report && receipt.checks.status === "failed", highFindings: high,
+    // The approved check as every surface reads it: a recorded pass the saved receipt can't show doesn't count.
+    const checks = receiptChecksOf(receipt);
+    const backing = checkBackingOf(checks);
+    const verdict = shownVerdictOf(receipt.proof?.verdict ?? null, { checkFailed: !report && checks?.status === "failed", highFindings: high,
         reviewPending: review?.state === "pending", criteriaPassed: matrix.every(row => requirementWordOf(row, null, backing) === "Met") });
-    const blockers = completionBlockersOf({ report, checks: receipt.checks, checkRequired, matrix, verdict: receipt.proof?.verdict ?? null,
+    const blockers = completionBlockersOf({ report, checks, checkRequired, matrix, verdict: receipt.proof?.verdict ?? null,
         accepted: receipt.proofAcceptance !== null, high });
     return { verdict, checksOff: !report && receipt.checks.level === "off", blockers };
 }