@plune-ai/cli 0.7.0 → 0.9.0

package/dist/reporter-core.cjs

@@ -128,12 +128,7 @@ function createClient(opts) {
 }
 
 // src/reporter-core/env.ts
-var UNSUPPORTED_VARS = [
-  "PLUNE_RUN_TITLE",
-  "PLUNE_ENV",
-  "PLUNE_LABELS",
-  "PLUNE_GROUP"
-];
+var UNSUPPORTED_VARS = ["PLUNE_GROUP"];
 function value(env, name) {
   const raw = env[name]?.trim();
   return raw === void 0 || raw === "" ? void 0 : raw;
@@ -148,19 +143,35 @@ function count(env, name) {
   const n = Number(raw);
   return Number.isInteger(n) && n > 0 ? n : void 0;
 }
+function list(env, name) {
+  const raw = value(env, name);
+  if (raw === void 0) return void 0;
+  const items = raw.split(",").map((s) => s.trim()).filter((s) => s !== "");
+  return items.length === 0 ? void 0 : items;
+}
 function readEnv(env = process.env) {
   const externalKey = value(env, "PLUNE_RUN");
   const token = value(env, "PLUNE_TOKEN");
   const apiUrl = value(env, "PLUNE_API_URL");
   const fallbackPath = value(env, "PLUNE_FALLBACK");
   const batchSize = count(env, "PLUNE_BATCH_SIZE");
+  const title = value(env, "PLUNE_RUN_TITLE");
+  const environment = value(env, "PLUNE_ENV");
+  const labels = list(env, "PLUNE_LABELS");
+  const offerDiscovered = flag(env, "PLUNE_CREATE");
   return {
     config: {
       ...externalKey !== void 0 ? { externalKey } : {},
       ...token !== void 0 ? { token } : {},
       ...apiUrl !== void 0 ? { apiUrl } : {},
       ...fallbackPath !== void 0 ? { fallbackPath } : {},
-      ...batchSize !== void 0 ? { batchSize } : {}
+      ...batchSize !== void 0 ? { batchSize } : {},
+      ...title !== void 0 ? { title } : {},
+      ...environment !== void 0 ? { environment } : {},
+      ...labels !== void 0 ? { labels } : {},
+      // Only when set: `false` here would outrank a config file that asked for it, and an unset
+      // variable is not an instruction.
+      ...offerDiscovered ? { offerDiscovered } : {}
     },
     // Two names for two situations that end the same way. `SHARED_RUN` says other processes are
     // reporting into this run; `PROCEED` says the job will close it on its own schedule. Either
@@ -189,6 +200,7 @@ function appendBatch(file, batch) {
 // src/reporter-core/session.ts
 var RESOLVE_CHUNK = 500;
 var BATCH_MAX = 500;
+var DISCOVER_MAX = 500;
 var BATCH_DEFAULT = 100;
 var EXPECTED_MAX = 1e4;
 function chunk(items, size) {
@@ -196,6 +208,14 @@ function chunk(items, size) {
   for (let i = 0; i < items.length; i += size) out.push(items.slice(i, i + size));
   return out;
 }
+function warnIfDropped(sent, run, log) {
+  const names = Object.keys(sent);
+  if (names.length === 0) return;
+  if (names.some((name) => run[name] != null)) return;
+  log(
+    `plune: this deployment stored none of ${names.join(", ")} \u2014 it is older than the fields. The run is otherwise reported in full.`
+  );
+}
 async function startRun(passed, expected) {
   const env = readEnv();
   const stated = Object.fromEntries(Object.entries(passed).filter(([, v]) => v !== void 0));
@@ -205,6 +225,7 @@ async function startRun(passed, expected) {
   const fallbackPath = cfg.fallbackPath ?? DEFAULT_FALLBACK_PATH;
   const externalKey = cfg.externalKey ?? null;
   const batchSize = Math.min(cfg.batchSize ?? BATCH_DEFAULT, BATCH_MAX);
+  const offerDiscovered = cfg.offerDiscovered ?? false;
   const token = cfg.token ?? loadToken() ?? "";
   const client = createClient({
     apiUrl: resolveApiUrl(cfg.apiUrl),
@@ -217,10 +238,12 @@ async function startRun(passed, expected) {
     conflict: 0,
     rejected: 0,
     unresolved: 0,
+    offered: 0,
     deferred: 0
   };
   const resolved = /* @__PURE__ */ new Map();
   let buffer = [];
+  const discoveries = /* @__PURE__ */ new Map();
   let runId = null;
   let joined = false;
   let offline = token === "";
@@ -263,16 +286,23 @@ async function startRun(passed, expected) {
     if (expected !== void 0 && expected.length > EXPECTED_MAX) {
       log(`plune: ${expected.length} tests is more than the run configuration holds \u2014 reporting the first ${EXPECTED_MAX}.`);
     }
+    const described = {
+      ...cfg.title !== void 0 ? { title: cfg.title } : {},
+      ...cfg.environment !== void 0 ? { environment: cfg.environment } : {},
+      ...cfg.labels !== void 0 ? { labels: cfg.labels } : {}
+    };
     const start = await client.post("/v1/runs", {
       schemaVersion: 2,
       kind: cfg.kind ?? "automated",
       ...externalKey !== null ? { externalKey } : {},
       ...cfg.meta !== void 0 ? { meta: cfg.meta } : {},
-      ...configuration !== void 0 ? { configuration } : {}
+      ...configuration !== void 0 ? { configuration } : {},
+      ...described
     });
     if (start.ok) {
       runId = start.body.run.id;
       joined = start.status === 200;
+      warnIfDropped(described, start.body.run, log);
     } else {
       log(`plune: could not start the run (${start.detail || start.kind}). Results will be written to ${fallbackPath}.`);
     }
@@ -287,6 +317,32 @@ async function startRun(passed, expected) {
     }
     return unanswered ? { kind: "unanswered" } : { kind: "unmatched" };
   }
+  function remember(result) {
+    if (!offerDiscovered) return;
+    if (result.title === void 0 || result.specRef === void 0) return;
+    if (result.keys.length === 0) return;
+    discoveries.set(result.keys[0].value, {
+      keys: result.keys,
+      title: result.title,
+      source: result.source,
+      specRef: result.specRef,
+      rawStatus: result.rawStatus
+    });
+  }
+  async function offer() {
+    if (discoveries.size === 0 || offline || runId === null) return;
+    for (const batch of chunk([...discoveries.values()], DISCOVER_MAX)) {
+      const out = await client.post("/v1/review-items/discovered", {
+        discovered: batch,
+        runId
+      });
+      if (!out.ok) {
+        log(`plune: could not offer ${batch.length} unknown tests for review (${out.detail || out.kind}).`);
+        return;
+      }
+      stats.offered += out.body.results.filter((r) => r.outcome === "queued").length;
+    }
+  }
   async function flush() {
     if (buffer.length === 0) return;
     const pending = buffer;
@@ -305,6 +361,7 @@ async function startRun(passed, expected) {
         submissions.push({ ...rest, testCaseId: match.testCaseId });
       } else if (match.kind === "unmatched") {
         stats.unresolved += 1;
+        remember(result);
       } else {
         unanswered.push(result);
       }
@@ -338,6 +395,7 @@ async function startRun(passed, expected) {
     if (stats.conflict > 0) parts.push(`${stats.conflict} conflicting`);
     if (stats.rejected > 0) parts.push(`${stats.rejected} rejected`);
     if (stats.unresolved > 0) parts.push(`${stats.unresolved} with no matching test case`);
+    if (stats.offered > 0) parts.push(`${stats.offered} offered for review`);
     if (stats.deferred > 0) parts.push(`${stats.deferred} written to ${fallbackPath}`);
     log(`plune: ${parts.join(" \xB7 ")}`);
   }
@@ -360,6 +418,7 @@ async function startRun(passed, expected) {
       if (done) return;
       done = true;
       await flush();
+      await offer();
       if (runId !== null && !offline) {
         const out = await client.post(`/v1/runs/${runId}/events`, {
           event: "finish",
@@ -373,6 +432,7 @@ async function startRun(passed, expected) {
       if (done) return;
       done = true;
       await flush();
+      await offer();
       if (runId !== null) {
         log(`plune: run ${runId} is still open \u2014 close it with "plune run finish ${runId}".`);
       }

package/dist/reporter-core.d.cts

@@ -88,6 +88,19 @@ interface PendingResult {
     rawStatus: string;
     /** What the author expected — `test.fail()` and friends. Absent means a pass was expected. */
     expectedStatus?: ResultStatus;
+    /**
+     * What to call this test, and where it lives — read ONLY when nothing resolved it (D14).
+     *
+     * A result that found its case needs neither: the case already has a title and a spec. These are
+     * for the other path, where the reporter offers the test to the review queue and the queue needs a
+     * name a person can judge and a spec a case can point at.
+     *
+     * Optional because an adapter that cannot say both simply does not participate. Deriving them from
+     * a key's internal spelling was the alternative and it is worse: it makes the core parse a string
+     * the adapter formatted, which is a contract nothing checks.
+     */
+    title?: string;
+    specRef?: string;
     execution?: Execution;
     errorContext?: string;
     params?: Record<string, unknown>;
@@ -121,8 +134,11 @@ interface SubmitCounts {
 /** What the session accumulated, for the one summary line a reporter prints. */
 interface RunStats extends SubmitCounts {
     /** Results whose keys matched no case — never sent, because inventing an id would write into
-     * somebody else's history. What to do about them is C2's ladder. */
+     * somebody else's history. Rung 6 of C2's ladder is what to do about them: see `offered`. */
     unresolved: number;
+    /** Unresolved tests offered to the review queue under `PLUNE_CREATE=1` (D14). Counted separately
+     * from `unresolved` because they are the same tests seen twice, not a second population. */
+    offered: number;
     /** Results written to the fallback file instead of the platform. */
     deferred: number;
 }
@@ -136,10 +152,24 @@ interface ReporterConfig {
     externalKey?: string;
     kind?: RunKind;
     meta?: RunMeta;
+    /** What to call this run in a list. The first shard to declare one wins. */
+    title?: string;
+    /** Where it ran — staging, prod, a preview. What keeps two CI matrices from reading as one. */
+    environment?: string;
+    /** How the run is marked — smoke, nightly, a team. */
+    labels?: string[];
     /** Results per request. The platform's ceiling is 500. */
     batchSize?: number;
     /** Where unsent batches go. Defaults beside the run file the CLI already writes. */
     fallbackPath?: string;
+    /**
+     * Offer tests that resolved to nothing to the platform's review queue (`PLUNE_CREATE=1`, D14).
+     *
+     * Off by default, and that is the product decision rather than caution: a reporter that filled a
+     * stranger's review queue on first run would teach the team to ignore the queue. Turning it on is
+     * how somebody says "yes, tell me what we are running that you do not know about".
+     */
+    offerDiscovered?: boolean;
     /** Transport seam — the same one `sync.ts` uses so tests need no server. */
     fetchImpl?: typeof fetch;
     /** Where a line for the operator goes. Defaults to stderr. */
@@ -167,15 +197,6 @@ interface RunSession {
     /** Send what is left and deliberately leave the run open — the shard case (AC-03). */
     leaveOpen(): Promise<void>;
 }
-/**
- * Start a run, or join the one that already carries this key.
- *
- * `expected` — one list of candidate identifiers per test the runner knows about — is used twice
- * on purpose. It becomes `configuration.expected`, from which the platform derives what never ran
- * (AC-09); and it is the input to the single up-front resolve, so the per-test path afterwards is
- * a map lookup rather than a request (AC-10). A runner that cannot enumerate its tests passes
- * nothing and resolves lazily on the first flush — same function, later moment.
- */
 declare function startRun(passed: ReporterConfig, expected?: readonly (readonly KeyRef[])[]): Promise<RunSession>;
 
 /**
@@ -209,14 +230,17 @@ declare function resultKey(testId: string, retry: number): string;
  * laptop the config was written on and the runner it executes on.
  */
 /**
- * Names the reporter promises but the platform cannot yet store (feature D13).
+ * Names the reporter promises but the platform cannot yet store.
  *
  * They are listed rather than ignored because of how they would fail otherwise: a run's `meta`
- * strips keys it does not know, so sending `PLUNE_ENV` there would be accepted, dropped, and look
- * exactly like success. A client believing it saved something the server discarded is the most
- * expensive kind of quiet — cheaper to say "not yet" out loud.
+ * strips keys it does not know, so sending one there would be accepted, dropped, and look exactly
+ * like success. A client believing it saved something the server discarded is the most expensive
+ * kind of quiet — cheaper to say "not yet" out loud.
+ *
+ * `PLUNE_RUN_TITLE`, `PLUNE_ENV` and `PLUNE_LABELS` left this list when D13 gave a run somewhere to
+ * put them. `PLUNE_GROUP` stays: a run group is D6's, and a group of one run means nothing.
  */
-declare const UNSUPPORTED_VARS: readonly ["PLUNE_RUN_TITLE", "PLUNE_ENV", "PLUNE_LABELS", "PLUNE_GROUP"];
+declare const UNSUPPORTED_VARS: readonly ["PLUNE_GROUP"];
 interface EnvSettings {
     /** Config fields the environment supplied. Anything passed explicitly outranks these. */
     config: Partial<ReporterConfig>;

package/dist/reporter-core.d.ts

@@ -88,6 +88,19 @@ interface PendingResult {
     rawStatus: string;
     /** What the author expected — `test.fail()` and friends. Absent means a pass was expected. */
     expectedStatus?: ResultStatus;
+    /**
+     * What to call this test, and where it lives — read ONLY when nothing resolved it (D14).
+     *
+     * A result that found its case needs neither: the case already has a title and a spec. These are
+     * for the other path, where the reporter offers the test to the review queue and the queue needs a
+     * name a person can judge and a spec a case can point at.
+     *
+     * Optional because an adapter that cannot say both simply does not participate. Deriving them from
+     * a key's internal spelling was the alternative and it is worse: it makes the core parse a string
+     * the adapter formatted, which is a contract nothing checks.
+     */
+    title?: string;
+    specRef?: string;
     execution?: Execution;
     errorContext?: string;
     params?: Record<string, unknown>;
@@ -121,8 +134,11 @@ interface SubmitCounts {
 /** What the session accumulated, for the one summary line a reporter prints. */
 interface RunStats extends SubmitCounts {
     /** Results whose keys matched no case — never sent, because inventing an id would write into
-     * somebody else's history. What to do about them is C2's ladder. */
+     * somebody else's history. Rung 6 of C2's ladder is what to do about them: see `offered`. */
     unresolved: number;
+    /** Unresolved tests offered to the review queue under `PLUNE_CREATE=1` (D14). Counted separately
+     * from `unresolved` because they are the same tests seen twice, not a second population. */
+    offered: number;
     /** Results written to the fallback file instead of the platform. */
     deferred: number;
 }
@@ -136,10 +152,24 @@ interface ReporterConfig {
     externalKey?: string;
     kind?: RunKind;
     meta?: RunMeta;
+    /** What to call this run in a list. The first shard to declare one wins. */
+    title?: string;
+    /** Where it ran — staging, prod, a preview. What keeps two CI matrices from reading as one. */
+    environment?: string;
+    /** How the run is marked — smoke, nightly, a team. */
+    labels?: string[];
     /** Results per request. The platform's ceiling is 500. */
     batchSize?: number;
     /** Where unsent batches go. Defaults beside the run file the CLI already writes. */
     fallbackPath?: string;
+    /**
+     * Offer tests that resolved to nothing to the platform's review queue (`PLUNE_CREATE=1`, D14).
+     *
+     * Off by default, and that is the product decision rather than caution: a reporter that filled a
+     * stranger's review queue on first run would teach the team to ignore the queue. Turning it on is
+     * how somebody says "yes, tell me what we are running that you do not know about".
+     */
+    offerDiscovered?: boolean;
     /** Transport seam — the same one `sync.ts` uses so tests need no server. */
     fetchImpl?: typeof fetch;
     /** Where a line for the operator goes. Defaults to stderr. */
@@ -167,15 +197,6 @@ interface RunSession {
     /** Send what is left and deliberately leave the run open — the shard case (AC-03). */
     leaveOpen(): Promise<void>;
 }
-/**
- * Start a run, or join the one that already carries this key.
- *
- * `expected` — one list of candidate identifiers per test the runner knows about — is used twice
- * on purpose. It becomes `configuration.expected`, from which the platform derives what never ran
- * (AC-09); and it is the input to the single up-front resolve, so the per-test path afterwards is
- * a map lookup rather than a request (AC-10). A runner that cannot enumerate its tests passes
- * nothing and resolves lazily on the first flush — same function, later moment.
- */
 declare function startRun(passed: ReporterConfig, expected?: readonly (readonly KeyRef[])[]): Promise<RunSession>;
 
 /**
@@ -209,14 +230,17 @@ declare function resultKey(testId: string, retry: number): string;
  * laptop the config was written on and the runner it executes on.
  */
 /**
- * Names the reporter promises but the platform cannot yet store (feature D13).
+ * Names the reporter promises but the platform cannot yet store.
  *
  * They are listed rather than ignored because of how they would fail otherwise: a run's `meta`
- * strips keys it does not know, so sending `PLUNE_ENV` there would be accepted, dropped, and look
- * exactly like success. A client believing it saved something the server discarded is the most
- * expensive kind of quiet — cheaper to say "not yet" out loud.
+ * strips keys it does not know, so sending one there would be accepted, dropped, and look exactly
+ * like success. A client believing it saved something the server discarded is the most expensive
+ * kind of quiet — cheaper to say "not yet" out loud.
+ *
+ * `PLUNE_RUN_TITLE`, `PLUNE_ENV` and `PLUNE_LABELS` left this list when D13 gave a run somewhere to
+ * put them. `PLUNE_GROUP` stays: a run group is D6's, and a group of one run means nothing.
  */
-declare const UNSUPPORTED_VARS: readonly ["PLUNE_RUN_TITLE", "PLUNE_ENV", "PLUNE_LABELS", "PLUNE_GROUP"];
+declare const UNSUPPORTED_VARS: readonly ["PLUNE_GROUP"];
 interface EnvSettings {
     /** Config fields the environment supplied. Anything passed explicitly outranks these. */
     config: Partial<ReporterConfig>;

package/dist/reporter-core.js

@@ -85,12 +85,7 @@ function createClient(opts) {
 }
 
 // src/reporter-core/env.ts
-var UNSUPPORTED_VARS = [
-  "PLUNE_RUN_TITLE",
-  "PLUNE_ENV",
-  "PLUNE_LABELS",
-  "PLUNE_GROUP"
-];
+var UNSUPPORTED_VARS = ["PLUNE_GROUP"];
 function value(env, name) {
   const raw = env[name]?.trim();
   return raw === void 0 || raw === "" ? void 0 : raw;
@@ -105,19 +100,35 @@ function count(env, name) {
   const n = Number(raw);
   return Number.isInteger(n) && n > 0 ? n : void 0;
 }
+function list(env, name) {
+  const raw = value(env, name);
+  if (raw === void 0) return void 0;
+  const items = raw.split(",").map((s) => s.trim()).filter((s) => s !== "");
+  return items.length === 0 ? void 0 : items;
+}
 function readEnv(env = process.env) {
   const externalKey = value(env, "PLUNE_RUN");
   const token = value(env, "PLUNE_TOKEN");
   const apiUrl = value(env, "PLUNE_API_URL");
   const fallbackPath = value(env, "PLUNE_FALLBACK");
   const batchSize = count(env, "PLUNE_BATCH_SIZE");
+  const title = value(env, "PLUNE_RUN_TITLE");
+  const environment = value(env, "PLUNE_ENV");
+  const labels = list(env, "PLUNE_LABELS");
+  const offerDiscovered = flag(env, "PLUNE_CREATE");
   return {
     config: {
       ...externalKey !== void 0 ? { externalKey } : {},
       ...token !== void 0 ? { token } : {},
       ...apiUrl !== void 0 ? { apiUrl } : {},
       ...fallbackPath !== void 0 ? { fallbackPath } : {},
-      ...batchSize !== void 0 ? { batchSize } : {}
+      ...batchSize !== void 0 ? { batchSize } : {},
+      ...title !== void 0 ? { title } : {},
+      ...environment !== void 0 ? { environment } : {},
+      ...labels !== void 0 ? { labels } : {},
+      // Only when set: `false` here would outrank a config file that asked for it, and an unset
+      // variable is not an instruction.
+      ...offerDiscovered ? { offerDiscovered } : {}
     },
     // Two names for two situations that end the same way. `SHARED_RUN` says other processes are
     // reporting into this run; `PROCEED` says the job will close it on its own schedule. Either
@@ -146,6 +157,7 @@ function appendBatch(file, batch) {
 // src/reporter-core/session.ts
 var RESOLVE_CHUNK = 500;
 var BATCH_MAX = 500;
+var DISCOVER_MAX = 500;
 var BATCH_DEFAULT = 100;
 var EXPECTED_MAX = 1e4;
 function chunk(items, size) {
@@ -153,6 +165,14 @@ function chunk(items, size) {
   for (let i = 0; i < items.length; i += size) out.push(items.slice(i, i + size));
   return out;
 }
+function warnIfDropped(sent, run, log) {
+  const names = Object.keys(sent);
+  if (names.length === 0) return;
+  if (names.some((name) => run[name] != null)) return;
+  log(
+    `plune: this deployment stored none of ${names.join(", ")} \u2014 it is older than the fields. The run is otherwise reported in full.`
+  );
+}
 async function startRun(passed, expected) {
   const env = readEnv();
   const stated = Object.fromEntries(Object.entries(passed).filter(([, v]) => v !== void 0));
@@ -162,6 +182,7 @@ async function startRun(passed, expected) {
   const fallbackPath = cfg.fallbackPath ?? DEFAULT_FALLBACK_PATH;
   const externalKey = cfg.externalKey ?? null;
   const batchSize = Math.min(cfg.batchSize ?? BATCH_DEFAULT, BATCH_MAX);
+  const offerDiscovered = cfg.offerDiscovered ?? false;
   const token = cfg.token ?? loadToken() ?? "";
   const client = createClient({
     apiUrl: resolveApiUrl(cfg.apiUrl),
@@ -174,10 +195,12 @@ async function startRun(passed, expected) {
     conflict: 0,
     rejected: 0,
     unresolved: 0,
+    offered: 0,
     deferred: 0
   };
   const resolved = /* @__PURE__ */ new Map();
   let buffer = [];
+  const discoveries = /* @__PURE__ */ new Map();
   let runId = null;
   let joined = false;
   let offline = token === "";
@@ -220,16 +243,23 @@ async function startRun(passed, expected) {
     if (expected !== void 0 && expected.length > EXPECTED_MAX) {
       log(`plune: ${expected.length} tests is more than the run configuration holds \u2014 reporting the first ${EXPECTED_MAX}.`);
     }
+    const described = {
+      ...cfg.title !== void 0 ? { title: cfg.title } : {},
+      ...cfg.environment !== void 0 ? { environment: cfg.environment } : {},
+      ...cfg.labels !== void 0 ? { labels: cfg.labels } : {}
+    };
     const start = await client.post("/v1/runs", {
       schemaVersion: 2,
       kind: cfg.kind ?? "automated",
       ...externalKey !== null ? { externalKey } : {},
       ...cfg.meta !== void 0 ? { meta: cfg.meta } : {},
-      ...configuration !== void 0 ? { configuration } : {}
+      ...configuration !== void 0 ? { configuration } : {},
+      ...described
     });
     if (start.ok) {
       runId = start.body.run.id;
       joined = start.status === 200;
+      warnIfDropped(described, start.body.run, log);
     } else {
       log(`plune: could not start the run (${start.detail || start.kind}). Results will be written to ${fallbackPath}.`);
     }
@@ -244,6 +274,32 @@ async function startRun(passed, expected) {
     }
     return unanswered ? { kind: "unanswered" } : { kind: "unmatched" };
   }
+  function remember(result) {
+    if (!offerDiscovered) return;
+    if (result.title === void 0 || result.specRef === void 0) return;
+    if (result.keys.length === 0) return;
+    discoveries.set(result.keys[0].value, {
+      keys: result.keys,
+      title: result.title,
+      source: result.source,
+      specRef: result.specRef,
+      rawStatus: result.rawStatus
+    });
+  }
+  async function offer() {
+    if (discoveries.size === 0 || offline || runId === null) return;
+    for (const batch of chunk([...discoveries.values()], DISCOVER_MAX)) {
+      const out = await client.post("/v1/review-items/discovered", {
+        discovered: batch,
+        runId
+      });
+      if (!out.ok) {
+        log(`plune: could not offer ${batch.length} unknown tests for review (${out.detail || out.kind}).`);
+        return;
+      }
+      stats.offered += out.body.results.filter((r) => r.outcome === "queued").length;
+    }
+  }
   async function flush() {
     if (buffer.length === 0) return;
     const pending = buffer;
@@ -262,6 +318,7 @@ async function startRun(passed, expected) {
         submissions.push({ ...rest, testCaseId: match.testCaseId });
       } else if (match.kind === "unmatched") {
         stats.unresolved += 1;
+        remember(result);
       } else {
         unanswered.push(result);
       }
@@ -295,6 +352,7 @@ async function startRun(passed, expected) {
     if (stats.conflict > 0) parts.push(`${stats.conflict} conflicting`);
     if (stats.rejected > 0) parts.push(`${stats.rejected} rejected`);
     if (stats.unresolved > 0) parts.push(`${stats.unresolved} with no matching test case`);
+    if (stats.offered > 0) parts.push(`${stats.offered} offered for review`);
     if (stats.deferred > 0) parts.push(`${stats.deferred} written to ${fallbackPath}`);
     log(`plune: ${parts.join(" \xB7 ")}`);
   }
@@ -317,6 +375,7 @@ async function startRun(passed, expected) {
       if (done) return;
       done = true;
       await flush();
+      await offer();
       if (runId !== null && !offline) {
         const out = await client.post(`/v1/runs/${runId}/events`, {
           event: "finish",
@@ -330,6 +389,7 @@ async function startRun(passed, expected) {
       if (done) return;
       done = true;
       await flush();
+      await offer();
       if (runId !== null) {
         log(`plune: run ${runId} is still open \u2014 close it with "plune run finish ${runId}".`);
       }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@plune-ai/cli",
-  "version": "0.7.0",
+  "version": "0.9.0",
   "description": "AI-powered assertion testing for LLM apps — write assertions, run them against any provider, get a pass/fail report (local, CI, or diff).",
   "keywords": [
     "ai",