@runcofactor/cli 0.55.0 → 0.56.0

package/dist/{scenarios-W7YNLBVO.js → scenarios-TU656T2B.js}

@@ -1,17 +1,17 @@
 import { createRequire as __cr } from "module"; import { fileURLToPath as __fup } from "url"; import { dirname as __dn } from "path"; const require = __cr(import.meta.url); const __filename = __fup(import.meta.url); const __dirname = __dn(__filename);
 import {
   apiRequest
-} from "./chunk-WWUEKQLR.js";
+} from "./chunk-EWK44F6V.js";
 import "./chunk-S3Q7MDCS.js";
 import {
   getArgValue,
   hasFlag,
   resolveConfig
-} from "./chunk-BJHAVM7Z.js";
+} from "./chunk-4MFMPC5P.js";
 import {
   writeRawOutput
-} from "./chunk-KL2ARGH6.js";
-import "./chunk-U3MRXQZS.js";
+} from "./chunk-2OHYMT3A.js";
+import "./chunk-RHWPZKRV.js";
 import "./chunk-TOWZAG3I.js";
 
 // src/scenarios.ts
@@ -279,4 +279,4 @@ async function handleReject(argv, apiUrl, token) {
 export {
   runScenarios
 };
-//# sourceMappingURL=scenarios-W7YNLBVO.js.map
+//# sourceMappingURL=scenarios-TU656T2B.js.map

package/dist/{schedule-EDHQJUUZ.js → schedule-DXJI6SYO.js}

@@ -12,24 +12,24 @@ import {
   handleRuns,
   hasTimingFlags,
   updateRule
-} from "./chunk-YTXZUWMN.js";
-import "./chunk-XM3WTJ7Y.js";
-import "./chunk-XWJJYZQC.js";
+} from "./chunk-62EXS76C.js";
+import "./chunk-DEVD2WD2.js";
+import "./chunk-XHEVDABB.js";
 import "./chunk-DEK6B6LX.js";
 import {
   apiRequest,
   getScalarArgValue
-} from "./chunk-WWUEKQLR.js";
+} from "./chunk-EWK44F6V.js";
 import "./chunk-S3Q7MDCS.js";
 import {
   getArgValue,
   hasFlag,
   resolveConfig
-} from "./chunk-BJHAVM7Z.js";
+} from "./chunk-4MFMPC5P.js";
 import {
   writeRawOutput
-} from "./chunk-KL2ARGH6.js";
-import "./chunk-U3MRXQZS.js";
+} from "./chunk-2OHYMT3A.js";
+import "./chunk-RHWPZKRV.js";
 import "./chunk-TOWZAG3I.js";
 
 // src/schedule.ts
@@ -342,4 +342,4 @@ async function runSchedule(argv) {
 export {
   runSchedule
 };
-//# sourceMappingURL=schedule-EDHQJUUZ.js.map
+//# sourceMappingURL=schedule-DXJI6SYO.js.map

package/dist/{session-4N62APNQ.js → session-RUFTJHF4.js}

@@ -1,7 +1,7 @@
 import { createRequire as __cr } from "module"; import { fileURLToPath as __fup } from "url"; import { dirname as __dn } from "path"; const require = __cr(import.meta.url); const __filename = __fup(import.meta.url); const __dirname = __dn(__filename);
 import {
   printServedGuide
-} from "./chunk-QHRWXCON.js";
+} from "./chunk-O7B2IYCZ.js";
 import "./chunk-JHPFIFRB.js";
 import {
   INTERACTION_CAPTURE_SCRIPT
@@ -9,8 +9,8 @@ import {
 import {
   LocalBrowserHost,
   USER_PAUSED_ERROR
-} from "./chunk-SCQVBLNR.js";
-import "./chunk-33U56YGD.js";
+} from "./chunk-XHGF6VC6.js";
+import "./chunk-CWICLBLU.js";
 import {
   callTool,
   createSession,
@@ -30,7 +30,7 @@ import {
   PlaywrightClient,
   captureElementAtPoint,
   dispatchBrowserToolWithTimeout
-} from "./chunk-3KBYI6RR.js";
+} from "./chunk-H6ZMKNP7.js";
 import {
   apiRequest,
   credentialUsername,
@@ -41,7 +41,7 @@ import {
   getApplicationArg,
   selectCredential,
   uploadStorageState
-} from "./chunk-WWUEKQLR.js";
+} from "./chunk-EWK44F6V.js";
 import {
   attachCliDiagnosticMetadata
 } from "./chunk-W7CFLOEI.js";
@@ -52,16 +52,16 @@ import {
   getArgValue,
   hasFlag,
   resolveConfig
-} from "./chunk-BJHAVM7Z.js";
+} from "./chunk-4MFMPC5P.js";
 import {
   writeRawOutput
-} from "./chunk-KL2ARGH6.js";
+} from "./chunk-2OHYMT3A.js";
 import "./chunk-AK7WEGGB.js";
 import {
   consoleLogger
 } from "./chunk-5QHORHJT.js";
-import "./chunk-JQG3V5E3.js";
-import "./chunk-U3MRXQZS.js";
+import "./chunk-2E3STBHS.js";
+import "./chunk-RHWPZKRV.js";
 import "./chunk-TOWZAG3I.js";
 
 // src/session/index.ts
@@ -2880,4 +2880,4 @@ export {
   buildToolArgs,
   runSession
 };
-//# sourceMappingURL=session-4N62APNQ.js.map
+//# sourceMappingURL=session-RUFTJHF4.js.map

package/dist/{skill-XFGMNOVM.js → skill-ZK76GQ47.js}

@@ -7,12 +7,12 @@ import {
   isInstallPermissionError,
   isValidInstallName,
   resolveInstallDir
-} from "./chunk-EGWVL5XA.js";
+} from "./chunk-VMAR67FF.js";
 import {
   brandSkillContentFor,
   brandTextFor,
   command
-} from "./chunk-KL2ARGH6.js";
+} from "./chunk-2OHYMT3A.js";
 import "./chunk-TOWZAG3I.js";
 
 // src/skill.ts
@@ -329,14 +329,14 @@ Run \`cofactor proofs guide\` for the full procedure. Use with every user-visibl
    \`cofactor workflow run <workflowId> --sandbox <sandboxId> --stream --org <org>\`
    Iterate to passing; the run uses the claimed sandbox's own environment, so do not pass --environment. Keep the sandbox id; if you lose it, \`cofactor sandbox claims --org <org>\` lists your live claims, and \`cofactor sandbox destroy <sandboxId>\` tears one down when you are done. After every application or Workflow test change, commit/push application edits, sync, and rerun. Use the final passing run ID; keep the checkout clean. Exact tree/SHA requirements are in the guide.
 5. **Push and open a DRAFT PR.** Commit/push remaining changes; if application or Workflow tests changed, repeat step 4 first. Note PR number \`<n>\` for attach.
-6. **Number observable acceptance criteria.** Supply both \`--text\` and \`--detail\` for each new number; reuse them for additional evidence. Use the most specific node/assertion ID via optional \`--node <id>\` and \`--asserted "Asserted that \u2026"\`. A performance run can prove a criterion: run it at the pushed head SHA, then attach it with \`--perf-run <prun_...>\`. One criterion can mix kinds and carry several items (a query change: a workflow run for behavior and a perf run for speed). An agent eval run proves a criterion the same way: a full, passing run at the pushed head SHA, attached with \`--eval-run <aerun_...>\` (any number per criterion). Every item must pass.
+6. **Number observable acceptance criteria.** Write each criterion, \`--text\` and \`--asserted\` in plain product language a reviewer understands: what the user sees or the product does, with no field names, column names, flags, IDs, UUIDs, SHAs or code identifiers (bad: "Asserted that an API-launched external run reads back triageEnabled=false"; good: "Asserted that an external run launched with triage turned off stays off, and defaults to on when not specified"). A criterion is one short observable outcome a user can see ("An external run started with triage off stays off"). Supply \`--text\` for each new number; reuse it for additional evidence; \`--detail\` is no longer needed. Use the most specific node/assertion ID via optional \`--node <id>\` and \`--asserted "Asserted that \u2026"\`. A performance run can prove a criterion: run it at the pushed head SHA, then attach it with \`--perf-run <prun_...>\`. One criterion can mix kinds and carry several items (a query change: a workflow run for behavior and a perf run for speed). An agent eval run proves a criterion the same way: a full, passing run at the pushed head SHA, attached with \`--eval-run <aerun_...>\` (any number per criterion). Every item must pass.
 7. **Attach after the draft PR exists.** Use exactly one subject flag set per command. For E2E evidence, omit optional \`--node\` only if no specific ID exists:
-   \`cofactor proofs attach --pr <n> --application <app> --criterion <n> --text "<criterion>" --detail "<why this matters>" --test <wf_...> --run <wrun_...> [--node <most-specific-id>] --asserted "Asserted that \u2026" --org <org>\`
+   \`cofactor proofs attach --pr <n> --application <app> --criterion <n> --text "<criterion>" --test <wf_...> --run <wrun_...> [--node <most-specific-id>] --asserted "Asserted that \u2026" --org <org>\`
    For performance evidence:
    \`cofactor proofs attach --pr <n> --application <app> --criterion <n> --perf-run <prun_...> --asserted "Asserted that \u2026" --org <org>\`
    A test containing any \`visual_checkpoint\` node is a UI test, even when it also has other steps. Attaching it runs the same test at the PR base commit and the proof page shows each checkpoint's base, head and diff. Put checkpoints before steps that depend on the new behavior so the base run can reach them.
-   Only for a criterion Cofactor cannot exercise at all (desktop client, hardware, third-party console; never a failing or flaky test), attach manual evidence: \`cofactor proofs attach --pr <n> --application <app> --criterion <n> [--text "<criterion>" --detail "<why>"] --manual --notes <text|@file> [--artifact <https-url>]... --org <org>\`. Say what you did and why Cofactor could not prove it; notes and links are shown to every reader, so redact secrets and customer data and never link signed URLs. It passes, is marked manual, and never promotes; re-attaching replaces it, so repeat every \`--artifact\`.
-8. **Render, check and update.** Replace the PR proof block with all of \`cofactor proofs render --pr <n> --org <org>\`; paste it verbatim and never hand-format or add a prose "Proof notes" section that restates it. It leads with a stoplight per criterion (\u{1F7E2} passed, \u{1F7E1} stale, \u{1F534} failed, \u26AA manual) and names each test, how it ran and what it asserted, with links; it is a snapshot as of the commit in its header, so render and replace it again after every push, attach or detach. Reviewer context goes in \`--detail\`. Check with \`cofactor proofs status --repo <owner/repo> --pr <n> --org <org>\`. Detach a whole criterion with \`cofactor proofs detach --pr <n> --criterion <n> --org <org>\`, or one item with \`--test <wf_...>\` / \`--perf-run <prun_...>\` / \`--eval-run <aerun_...>\` / \`--manual\`; render again after every attach or detach.
+   Only for a criterion Cofactor cannot exercise at all (desktop client, hardware, third-party console; never a failing or flaky test), attach manual evidence (shown as checked another way): \`cofactor proofs attach --pr <n> --application <app> --criterion <n> [--text "<criterion>"] --manual --notes <text|@file> [--artifact <https-url>]... --org <org>\`. Say what you did and why Cofactor could not prove it; notes and links are shown to every reader, so redact secrets and customer data and never link signed URLs. It passes, is shown as checked another way, and never promotes; re-attaching replaces it, so repeat every \`--artifact\`. If you did not or could not check a criterion at all, say so: \`cofactor proofs attach --pr <n> --application <app> --criterion <n> --text "<criterion>" --not-proved "<why>" --org <org>\` (a warning, not a failure, in product language).
+8. **Render, check and update.** Replace the PR proof block with all of \`cofactor proofs render --pr <n> --org <org>\`; paste it verbatim and never hand-format or add a prose "Proof notes" section that restates it. It is one table with a stoplight per criterion (\u{1F7E2} proved by Cofactor, with stale as a note on it; \u26AA checked another way; \u{1F7E1} not proved, with your reason; \u{1F534} failed) and a row per evidence item naming the test, how it ran, what it asserted and a link; it is a snapshot as of the commit in its header, so render and replace it again after every push, attach or detach. Reviewer context goes in the \`Asserted that \u2026\` line. Check with \`cofactor proofs status --repo <owner/repo> --pr <n> --org <org>\`. Detach a whole criterion with \`cofactor proofs detach --pr <n> --criterion <n> --org <org>\`, or one item with \`--test <wf_...>\` / \`--perf-run <prun_...>\` / \`--eval-run <aerun_...>\` / \`--manual\`; render again after every attach or detach.
 
 Diagnose with \`cofactor proofs check --pr <n> --sha <head> --body-file <file> --org <org>\` or \`--body -\` for stdin; see the guide for checks/stale evidence. Re-run refreshes stale E2E evidence; performance evidence needs a new run at the current head, an attach, and a detach of the old item; manual evidence is refreshed by attaching it again at the new head. Server and CI own promotion triggers; \`cofactor trigger\` is not your job. No observable acceptance criterion at all (dependency bump, docs-only change)? Print this neutral block with \`cofactor proofs render --none "<why>"\` and paste it, replacing the PR proof block; never hand-type it.
 
@@ -1181,4 +1181,4 @@ export {
   resolveInstallDir,
   runSkill
 };
-//# sourceMappingURL=skill-XFGMNOVM.js.map
+//# sourceMappingURL=skill-ZK76GQ47.js.map