@logic-joe/fleek 0.2.12 → 0.2.14

package/dist/fleek.mjs

@@ -1055,7 +1055,8 @@ var require_claims = __commonJS({
   "../types/dist/agent-platform/claims.js"(exports) {
     "use strict";
     Object.defineProperty(exports, "__esModule", { value: true });
-    exports.CLAIM_QUOTAS = exports.AGENT_CONVENTION_DEFAULTS = exports.ANSWER_ASSURANCE_KINDS = exports.ISSUE_COMMENT_KINDS = exports.CLAIM_TTL_LIMITS = exports.CLAIM_END_REASONS_RETURNING_TO_POOL = exports.CLAIM_OUTCOME_END_REASON = exports.CLAIM_END_REASONS = exports.CLAIM_RELEASE_OUTCOMES = exports.CLAIM_COMPLETE_OUTCOMES = exports.CLAIM_PURPOSES = void 0;
+    exports.CLAIM_QUOTAS = exports.AGENT_CONVENTION_DEFAULTS = exports.ANSWER_ASSURANCE_KINDS = exports.ISSUE_COMMENT_KINDS = exports.CLAIM_TTL_LIMITS = exports.CLAIM_END_REASONS_RETURNING_TO_POOL = exports.WORK_DELIVERY_SHA = exports.WORK_DELIVERY_BRANCH = exports.WORK_DELIVERY_LIMITS = exports.CLAIM_STATE_KEPT_REASONS = exports.CLAIM_OUTCOME_END_REASON = exports.CLAIM_END_REASONS = exports.CLAIM_RELEASE_OUTCOMES = exports.CLAIM_COMPLETE_OUTCOMES = exports.CLAIM_PURPOSES = void 0;
+    exports.validDeliveryBranch = validDeliveryBranch2;
     exports.CLAIM_PURPOSES = ["work", "review"];
     exports.CLAIM_COMPLETE_OUTCOMES = ["done", "partial", "handed_off"];
     exports.CLAIM_RELEASE_OUTCOMES = ["paused", "not_doable"];
@@ -1076,6 +1077,21 @@ var require_claims = __commonJS({
       paused: "released",
       not_doable: "released"
     };
+    exports.CLAIM_STATE_KEPT_REASONS = [
+      "agents_may_complete_off",
+      "target_unavailable",
+      "rules_changed",
+      "review_not_allowed",
+      "state_changed",
+      "edit_right_missing",
+      "transition_denied"
+    ];
+    exports.WORK_DELIVERY_LIMITS = { branchChars: 200 };
+    exports.WORK_DELIVERY_BRANCH = /^[A-Za-z0-9][A-Za-z0-9._/-]{0,199}$/;
+    exports.WORK_DELIVERY_SHA = /^[0-9a-f]{40,64}$/;
+    function validDeliveryBranch2(name) {
+      return exports.WORK_DELIVERY_BRANCH.test(name) && !name.includes("..") && !name.includes("//") && !name.endsWith("/") && !name.endsWith(".lock");
+    }
     exports.CLAIM_END_REASONS_RETURNING_TO_POOL = ["released", "expired", "force_released", "owner_inactive", "revoked"];
     exports.CLAIM_TTL_LIMITS = { defaultMinutes: 60, minMinutes: 15, maxMinutes: 480, hardHours: 24 };
     exports.ISSUE_COMMENT_KINDS = ["comment", "progress", "question", "decision_request", "answer", "handoff", "result", "system"];
@@ -1610,7 +1626,9 @@ var require_error_codes = __commonJS({
       // sign-in on another Fleek host replaced the shared cookie) - the web client renews its session at once (selected by the
       // HttpExceptionFilter from a server-owned request mark, never by a thrower)
       "session_mismatch",
-      "public_visibility"
+      "public_visibility",
+      // MCP-FIX-M: a claim-bound agent write into a completed status while the project setting "Agents may complete" is off
+      "agents_may_complete_off"
     ];
     exports.WORKSPACE_APPS = ["chat", "projects", "documents", "whiteboards", "workflows"];
     exports.MAIL_DELIVERY_STATUSES = ["queued", "sending", "accepted", "ambiguous", "failed", "suppressed"];
@@ -1793,7 +1811,7 @@ var require_events = __commonJS({
       "issue.deleted": { inProcessEmitter: "issue.deleted", webhookWireEvent: "issue.deleted" }
     };
     exports.AGENT_EVENT_AGGREGATE_TYPES = ["issue", "approval", "workflow_run", "workflow_definition"];
-    exports.APPROVAL_EVENT_READER_METADATA = ["proposalId", "kind", "outcome"];
+    exports.APPROVAL_EVENT_READER_METADATA = ["proposalId", "approvalId", "kind", "outcome"];
     exports.DOMAIN_EVENT_VISIBILITIES = ["private", "internal", "public"];
     exports.APPROVAL_STALE_REASONS = [
       "revision_changed",
@@ -3686,8 +3704,9 @@ var require_client_ai = __commonJS({
       // AGD-62 LJK-1: the acceptance verdict of `fleek_complete_work` is LJ-internal (never in the guest toolset)
       "acceptance",
       "nextStep",
-      // ONF-200 KNOW-A: the doc statement of `fleek_complete_work` is LJ-internal too
-      "knowledge"
+      // ONF-200 KNOW-A: the doc statement of `fleek_complete_work` is LJ-internal too; PROTOCOL-NEXT-M: so is the delivered branch + sha
+      "knowledge",
+      "delivery"
     ];
     exports.CLIENT_AI_TOOLS = Object.freeze(Object.keys(exports.CLIENT_AI_TOOL_FIELDS).sort());
     function isClientAiTool(name) {
@@ -4633,7 +4652,7 @@ var require_setup_session = __commonJS({
     exports.SETUP_SESSION_TTL_MINUTES = 30;
     exports.SETUP_SESSION_OPEN_MAX = 10;
     exports.SETUP_SESSION_CLAIM_LIMIT = 10;
-    exports.FLEEK_CLI_SETUP_VERSION_DEFAULT = "0.2.12";
+    exports.FLEEK_CLI_SETUP_VERSION_DEFAULT = "0.2.14";
     exports.FLEEK_CLI_PACKAGE = "@logic-joe/fleek";
     exports.FLEEK_CANONICAL_SERVER = "https://api.onfleek.app";
     exports.SETUP_CLIENT_APPS = ["claude-code", "codex", "orca", "cursor", "other"];
@@ -4676,6 +4695,8 @@ var require_setup_session = __commonJS({
       "assign_kit",
       "switch_kit",
       "choose_repo",
+      "open_applications",
+      "check_repo_access",
       "show_command",
       "review_init",
       "setup_review_agent",
@@ -32703,6 +32724,8 @@ var require_knowledge_skeleton = __commonJS({
       ["agent_git_commit", "Agenten dürfen committen", "Agents may commit"],
       ["agent_git_push", "Agenten dürfen Feature-Branch pushen", "Agents may push a feature branch"],
       ["agent_git_open_mr", "Agenten dürfen MR öffnen", "Agents may open a merge request"],
+      // ONF-257: merging the own MR into the base (after green checks and the other model's clean review); empty / false = a person merges
+      ["agent_git_merge", "Agenten dürfen eigene MR mergen", "Agents may merge their own merge request"],
       ["agent_git_delete_local", "Agenten dürfen lokal aufräumen", "Agents may clean up locally"],
       ["agent_git_delete_remote", "Agenten dürfen Remote-Branch löschen", "Agents may delete the remote branch"],
       ["agent_git_stop_servers", "Agenten dürfen fremde Server stoppen", "Agents may stop other servers"]
@@ -33784,6 +33807,7 @@ var require_setup_proposal2 = __commonJS({
     exports.setupProposeShape = {
       packageId: uuid.optional(),
       revision: zod_1.z.number().int().min(1).optional(),
+      templateName: process_template_js_1.processTemplateName.optional(),
       entry: zod_1.z.enum(["example", "copy", "walkthrough"]),
       base: exports.setupBaseRefSchema,
       customer: zod_1.z.union([
@@ -33812,7 +33836,7 @@ var require_setup_proposal2 = __commonJS({
     exports.setupProposalStatusInputSchema = zod_1.z.object({ packageId: uuid }).strict();
     var edge = (rule) => `${rule.fromStateKey ?? "*"}>${rule.toStateKey}`;
     var ACTOR_RANK = { any_member: 0, lj_role: 1, named: 2, client: 3 };
-    function setupRisks(base, document3) {
+    function setupRisks(base, document3, invites = []) {
       const next = document3.process.document;
       const out = [];
       const seen = /* @__PURE__ */ new Set();
@@ -33899,7 +33923,7 @@ var require_setup_proposal2 = __commonJS({
         if (items(base).some((text6) => !now.includes(text6)))
           add({ code: "checklist_items_removed", params: { gate: role } });
       }
-      const productOwner = document3.people.some((person) => person.role === "product_owner");
+      const productOwner = !(0, setup_package_js_1.setupOpenPoints)(document3, invites).includes("no_product_owner");
       if (!productOwner && next.rules.some((rule) => rule.actorRequirement === "named" && rule.deciderRoles.includes("product_owner")))
         add({ code: "no_product_owner", params: {} });
       return out;
@@ -36095,9 +36119,19 @@ var require_agent_error_details = __commonJS({
     var transitionKeys = (value) => Array.isArray(value) && value.length <= 32 && value.every((item2) => typeof item2 === "string" && item2.length > 0 && item2.length <= 64 && /^[A-Za-z0-9_.:-]+$/.test(item2) && ([...types_1.TRANSITION_CONDITIONS, ...types_1.TRANSITION_REQUIRED_FIELDS].includes(item2) || /^[a-z0-9_.:-]+$/.test(item2)));
     var bindingField = (value) => typeof value === "string" && value.length <= 130 && (/^state:(?:[a-z][a-z0-9_]{0,31}|\+[0-9a-f-]{36})\.(?:role|gate|omitted)$/.test(value) || /^rule:\+[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/.test(value) || /^rule:(?:\*|[a-z][a-z0-9_]{0,31})>[a-z][a-z0-9_]{0,31}$/.test(value) || ["phases", "roles.defaultRole", "routeOptions", "clarification.maxQuestions", "clarification.maxQuestionsPerRound", "clarification.maxRounds", "clarification.answerWithinHours", "clarification.handoffRouteKey", "clarification.handoffUserId"].includes(value));
     var schema = {
+      APPROVAL_EXPIRED: {
+        id: uuid,
+        workspaceId: uuid,
+        requesterId: uuid,
+        toolName: (value) => typeof value === "string" && /^fleek_[a-z0-9_]{1,100}$/.test(value),
+        createdAt: (value) => typeof value === "string" && /^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{3}Z$/.test(value),
+        expiresAt: (value) => typeof value === "string" && /^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{3}Z$/.test(value),
+        outcome: oneOf3(["expired", "approved", "denied", "consumed"])
+      },
       TICKET_TEXT_LIMIT: {},
       TICKET_TEXT_FORMAT: {},
       TICKET_TEXT_LOSSY_REPLACEMENT: {},
+      NOT_FOUND: { reason: oneOf3(["no_setup"]) },
       ACCESS_DENIED: { reason: oneOf3(["approval_denied", "field_right", "variant_admin"]), field: bindingField, changeRequest },
       // ENGINE-AUTHORITY: engine setup refusals (`workflows.engine.*`) - closed lists only
       CONFLICT: {
@@ -36116,7 +36150,7 @@ var require_agent_error_details = __commonJS({
       // PRIVACY-WORKCTX: additive `holderRef` (person ref of the holder, null without a handle key / release owner)
       ALREADY_CLAIMED: { holderLabel: string4, expiresAt: string4, holderRef: personRef },
       CLAIMED_BY_OTHER: { holderLabel: string4, expiresAt: string4, holderRef: personRef },
-      CLAIM_REQUIRED: { issueIds: list2 },
+      CLAIM_REQUIRED: { issueIds: list2, requiredPurpose: oneOf3(["work"]), providedPurpose: oneOf3(["review"]) },
       RENEW_LIMIT: { hardExpiresAt: string4 },
       NOT_READY: { readySelector: string4 },
       REVISION_CONFLICT: { currentRevision: integer },
@@ -36259,6 +36293,8 @@ var require_agent_error_details = __commonJS({
           "repository",
           // AGD-62 LJK-1: the typed result's acceptance verdict (work.complete; explicit, not only through the guest refusal list)
           "acceptance",
+          // DEVX-START follow-up: the delivered branch + head commit of work.complete
+          "delivery",
           // KNOW-INGEST: run / submit / review inputs (names only)
           ...types_1.KNOWLEDGE_INGEST_INVALID_FIELDS,
           // AGENT-ATTACH: the file of `issues.attachFile` (names only, never the value)
@@ -36274,8 +36310,10 @@ var require_agent_error_details = __commonJS({
         ]),
         index: integer,
         path: (value) => (0, types_1.isProjectInitPath)(value) || (0, types_1.isKnowledgeSubmitPath)(value),
+        // MCP-FIX-M `handoff_target` = fleek_complete_work handoff names not exactly one of toUserId / toAgent;
+        // ONF-306 (F38): `page_not_internal` / `page_not_in_project_space` / `profile_schema_missing` = a profile page binding refused for a readable page;
         // WF-TF-2 `not_target_project`; WORK-FOLDER-b `human_only` = an agent assessment named an item only a human sets
-        reason: oneOf3(["release_name", ...types_1.WORK_REQUEST_INVALID_REASONS, "incompatible_work_filters", "not_target_project", "human_only", "system_item", ...types_1.SKILL_INVALID_REASONS, ...types_1.KIT_INVALID_INPUT_REASONS, ...types_1.SETUP_INVALID_REASONS, ...types_1.SETUP_PROPOSAL_INVALID_REASONS, ...types_1.KNOWLEDGE_SETUP_REASONS, ...types_1.PROJECT_INIT_INVALID_REASONS, ...types_1.KNOWLEDGE_INGEST_INVALID_REASONS]),
+        reason: oneOf3(["release_name", ...types_1.WORK_REQUEST_INVALID_REASONS, "incompatible_work_filters", "handoff_target", "page_not_internal", "page_not_in_project_space", "profile_schema_missing", "not_target_project", "human_only", "system_item", ...types_1.SKILL_INVALID_REASONS, ...types_1.KIT_INVALID_INPUT_REASONS, ...types_1.SETUP_INVALID_REASONS, ...types_1.SETUP_PROPOSAL_INVALID_REASONS, ...types_1.KNOWLEDGE_SETUP_REASONS, ...types_1.PROJECT_INIT_INVALID_REASONS, ...types_1.KNOWLEDGE_INGEST_INVALID_REASONS]),
         maxBytes: integer,
         maxItems: integer,
         messageKey: oneOf3(["agentWorkSkills.errors.fieldLimit", "agentWorkSkills.errors.countLimit"]),
@@ -36325,7 +36363,8 @@ var require_agent_error_details = __commonJS({
       // AGD-51: `openPoints` = the same closed key shape (condition / checklist keys; only for a proven unrestricted human)
       TRANSITION_DENIED: { reason: oneOf3(types_1.TRANSITION_DENIED_REASONS), toStateId: uuid, collectingStateId: uuid, missing: transitionKeys, openPoints: transitionKeys },
       TEAM_PROCESS_INVALID: { reason: oneOf3(types_1.TEAM_PROCESS_INVALID_REASONS) },
-      FIELD_LOCKED: { field: oneOf3([...types_1.TEAM_PROCESS_LOCKED_FIELDS, "projectReleaseId"]) },
+      // TES-403: `estimate` = the legacy point estimate a restricted actor never writes
+      FIELD_LOCKED: { field: oneOf3([...types_1.TEAM_PROCESS_LOCKED_FIELDS, "projectReleaseId", "estimate"]) },
       // AGD COST-LIMITS-a: the unit only (never an amount)
       COST_LIMIT_REACHED: { unit: oneOf3(["tokens", "eur"]) },
       // AGD TRIAGE-a: the project only
@@ -37337,6 +37376,7 @@ var EN = {
   "link.confirmStart": 'Start {key} "{title}" in Orca with {agent}? Task: {task}.',
   "errors.startForeignClaim": "{key} is being worked on by {holder} until {expires}. Nothing was started.",
   "errors.startReviewNoBranch": "{key}: origin has no branch with this ticket key, so nothing has been delivered for review. Nothing was started.",
+  "errors.startReviewMoved": "{key}: the delivered branch {branch} moved on since the delivery ({sha} delivered, origin now at {head}). Ask the builder to deliver again. Nothing was started.",
   "errors.startReviewBranches": "{key}: several branches carry this ticket key: {branches}. Nothing was started.",
   "errors.startReviewBranches.next": "fleek start {key} after removing or renaming the other branches",
   "errors.startReviewRefused": "{key}: the server does not allow a review claim right now.",
@@ -38103,6 +38143,7 @@ var DE = {
   "link.confirmStart": "{key} „{title}“ mit {agent} in Orca starten? Auftrag: {task}.",
   "errors.startForeignClaim": "{key} wird gerade von {holder} bearbeitet, bis {expires}. Es wurde nichts gestartet.",
   "errors.startReviewNoBranch": "{key}: Auf origin gibt es keinen Branch mit diesem Ticket-Key, also wurde nichts zum Review abgeliefert. Es wurde nichts gestartet.",
+  "errors.startReviewMoved": "{key}: Der abgelieferte Branch {branch} hat sich seit der Ablieferung weiterbewegt ({sha} abgeliefert, origin jetzt bei {head}). Bitte den Builder, neu abzuliefern. Es wurde nichts gestartet.",
   "errors.startReviewBranches": "{key}: Mehrere Branches tragen diesen Ticket-Key: {branches}. Es wurde nichts gestartet.",
   "errors.startReviewBranches.next": "fleek start {key}, nachdem die anderen Branches entfernt oder umbenannt sind",
   "errors.startReviewRefused": "{key}: Der Server erlaubt gerade keinen Review-Claim.",
@@ -40147,11 +40188,17 @@ function matchResponse(raw, id) {
   }
   return null;
 }
-function retrySafe(message) {
+function retrySafe(message, readOnlyTools = /* @__PURE__ */ new Set()) {
   if (message.method !== "tools/call") return true;
-  const args = object2(message.params) ? message.params.arguments : void 0;
+  const params = object2(message.params) ? message.params : void 0;
+  if (params && typeof params.name === "string" && readOnlyTools.has(params.name)) return true;
+  const args = params ? params.arguments : void 0;
   return object2(args) && typeof args.idempotencyKey === "string" && args.idempotencyKey.length > 0;
 }
+function readOnlyToolNames(answer2) {
+  const tools = object2(answer2.result) && Array.isArray(answer2.result.tools) ? answer2.result.tools : [];
+  return tools.filter((tool) => object2(tool) && typeof tool.name === "string" && object2(tool.annotations) && tool.annotations.readOnlyHint === true).map((tool) => tool.name);
+}
 var CODE = /^[A-Z_]{1,64}$/;
 var PROFILE_HINT = {
   PROFILE_AMBIGUOUS: "Several Fleek sign-ins are bound to this checkout: pin one with `--profile <key>` or FLEEK_PROFILE in the MCP entry.",
@@ -40174,6 +40221,7 @@ async function runProxy(options) {
   let session = null;
   let loginHinted = false;
   const reported = /* @__PURE__ */ new Set();
+  const readOnlyTools = /* @__PURE__ */ new Set();
   let closed = false;
   const write2 = (message) => {
     if (closed) return;
@@ -40272,6 +40320,7 @@ async function runProxy(options) {
         try {
           const answer2 = await post(message, id, signal);
           if (answer2 && message.method === "initialize" && object2(answer2.result) && typeof answer2.result.protocolVersion === "string" && PROXY_LEGACY_VERSIONS.includes(answer2.result.protocolVersion)) version2 = answer2.result.protocolVersion;
+          if (answer2 && message.method === "tools/list") for (const name of readOnlyToolNames(answer2)) readOnlyTools.add(name);
           if (answer2 && !signal.aborted) write2(answer2);
           if (answer2 && message.method === "initialize" && !object2(answer2.error)) loginHinted = false;
           return;
@@ -40287,7 +40336,7 @@ async function runProxy(options) {
             return;
           }
           const transient = error instanceof TransientError;
-          if (transient && attempt < retryDelays.length && retrySafe(message)) {
+          if (transient && attempt < retryDelays.length && retrySafe(message, readOnlyTools)) {
             await new Promise((resolve9) => setTimeout(resolve9, retryDelays[attempt]));
             if (signal.aborted) return;
             continue;
@@ -40299,7 +40348,7 @@ async function runProxy(options) {
             options.diagnostic(`fleek mcp proxy: request failed (${code})${PROFILE_HINT[code] ? ` - ${PROFILE_HINT[code]}` : ""}`);
           }
           if (id !== null) {
-            write2(transient ? proxyError(id, "UNAVAILABLE", retrySafe(message) ? "Fleek is unreachable right now; retry later." : "Fleek is unreachable right now; the outcome is unknown. Retry with an idempotencyKey to repeat it safely.") : proxyError(id, code, PROFILE_HINT[code] ?? "The request failed."));
+            write2(transient ? proxyError(id, "UNAVAILABLE", retrySafe(message, readOnlyTools) ? "Fleek is unreachable right now; retry later." : "Fleek is unreachable right now; the outcome is unknown. Retry with an idempotencyKey to repeat it safely.") : proxyError(id, code, PROFILE_HINT[code] ?? "The request failed."));
           }
           return;
         }
@@ -50082,13 +50131,12 @@ function causeOf(error) {
 
 // src/start.ts
 var AGENT = /^[a-z][a-z0-9-]{0,31}$/;
-var BRANCH = /^[A-Za-z0-9][A-Za-z0-9._/-]{0,200}$/;
-var validBranch = (name) => BRANCH.test(name) && !name.includes("..") && !name.includes("//") && !name.endsWith("/") && !name.endsWith(".lock");
+var validBranch = import_types6.validDeliveryBranch;
 var CANDIDATES_SHOWN = 10;
 var string3 = (value) => typeof value === "string" ? value : "";
 async function liveFacts(client, task) {
   const tools = await client.availableTools();
-  if (!tools.has(WIRE.context)) return { step: null, claims: [], providers: null };
+  if (!tools.has(WIRE.context)) return { step: null, claims: [], providers: null, delivery: null };
   const context = (await client.callTool(WIRE.context, { issueId: task.id, format: "json", include: ["task", "work", "conventions"] })).structured;
   const work2 = object3(context.work) ? context.work : {};
   const claims = (Array.isArray(work2.claims) ? work2.claims : []).filter(object3).map((claim) => ({
@@ -50098,6 +50146,8 @@ async function liveFacts(client, task) {
     holder: object3(claim.holder) ? string3(claim.holder.text) : "",
     expiresAt: string3(claim.expiresAt)
   }));
+  const recorded = object3(work2.delivery) ? work2.delivery : null;
+  const delivery = recorded && typeof recorded.branch === "string" && validBranch(recorded.branch) && typeof recorded.sha === "string" && import_types6.WORK_DELIVERY_SHA.test(recorded.sha) ? { branch: recorded.branch, sha: recorded.sha, at: string3(recorded.at) } : null;
   const conventions = object3(context.conventions) ? context.conventions : {};
   const providers = Array.isArray(conventions.aiProviders) ? conventions.aiProviders.filter((entry) => typeof entry === "string") : null;
   const taskSection = object3(context.task) ? context.task : {};
@@ -50107,7 +50157,7 @@ async function liveFacts(client, task) {
     const state = (await pages(client, "fleek_list_workflow_states", { teamId: task.projectId }, 200)).find((row) => string3(row.id).toLowerCase() === stateId);
     step = state && (0, import_types6.isOpenInStep)(state.role) ? state.role : null;
   }
-  return { step, claims, providers };
+  return { step, claims, providers, delivery };
 }
 function defaultAgent(providers) {
   return providers && providers.length && !providers.includes("anthropic") && providers.includes("openai") ? "codex" : "claude";
@@ -50162,7 +50212,14 @@ function branchesOfTicket(lsRemote, key) {
   const token = new RegExp(`(^|[^a-z0-9])${key.toLowerCase().replace(/[^a-z0-9-]/g, "")}($|[^a-z0-9])`, "i");
   return lsRemote.split("\n").map((line) => /^[0-9a-f]{40,64}\s+refs\/heads\/(\S+)$/.exec(line.trim())?.[1] ?? "").filter((name) => name && validBranch(name) && token.test(name));
 }
-async function deliveredBranch(deps, root, key) {
+async function deliveredBranch(deps, root, key, recorded) {
+  if (recorded) {
+    if ((await fetchBranch(deps, root, recorded.branch)).code !== 0) throw new CliError("errors.startGit", EXIT.failed, { key, path: neutralizeLine(root) }, "GIT_FAILED");
+    const head2 = /^([0-9a-f]{40,64})\s+(\S+)/.exec((await git(deps, root, ["log", "-1", "--format=%H %cI", `refs/remotes/origin/${recorded.branch}`])).stdout.trim());
+    if (!head2) throw new CliError("errors.startGit", EXIT.failed, { key, path: neutralizeLine(root) }, "GIT_FAILED");
+    if (head2[1] !== recorded.sha) throw new CliError("errors.startReviewMoved", EXIT.failed, { key, branch: neutralizeLine(recorded.branch), sha: recorded.sha.slice(0, 10), head: head2[1].slice(0, 10) }, "REVIEW_BRANCH_MOVED");
+    return { branch: recorded.branch, sha: recorded.sha, date: /^\d{4}-\d{2}-\d{2}/.test(recorded.at) ? recorded.at.slice(0, 10) : head2[2].slice(0, 10) };
+  }
   const remote = await git(deps, root, ["ls-remote", "--heads", "origin"]);
   if (remote.code !== 0) throw new CliError("errors.startGit", EXIT.failed, { key, path: neutralizeLine(root) }, "GIT_FAILED");
   const found = branchesOfTicket(remote.stdout, key);
@@ -50255,7 +50312,7 @@ async function runStart(deps, input) {
   let target = purpose === "work" ? existing[0] ?? null : null;
   if (purpose === "review") {
     try {
-      delivered = await deliveredBranch(deps, root, task.key);
+      delivered = await deliveredBranch(deps, root, task.key, live.delivery);
     } catch (error) {
       target = existing.find((entry) => entry.terminal) ?? null;
       if (!target) throw error;
@@ -51678,7 +51735,7 @@ var USER_SKILL_MARKER = "<!-- Generated by scripts/agent-guidance/generate.mjs f
 var USER_SKILLS = [
   {
     "name": "fleek-orca-coordinator",
-    "content": "---\nname: fleek-orca-coordinator\ndescription: \"Coordinate Fleek tickets in Orca: find claimable work, claim it for this machine, start a builder in its own worktree with the project kit, review with the other model, leave gates to people and deliver to review. Use when you coordinate Orca agents for a Fleek project.\"\nmetadata:\n  protocol: fleek-work-protocol/1.13\n---\n\n<!-- Generated by scripts/agent-guidance/generate.mjs from scripts/agent-guidance/protocol.json - do not edit by hand. -->\n\n# Fleek in Orca: coordinator recipe\n\nFleek holds the tickets and every human decision; Orca runs worktrees and agents. Follow the fleek-work skill for every Fleek call. Never change Orca's code or settings.\n\n## 1. Ready check\n- `fleek doctor` in the repo: Fleek MCP current for Claude Code, Codex and every Orca account, repo bound to its Fleek project. Otherwise ask the person to run the setup command from Fleek once per machine (Settings › Connections › Set up this computer: `fleek setup <code>`). Never sign in for them.\n\n## 2. Find work\n- fleek_find_work and fleek_my_work_inbox (claimable: agent-ready or assigned to your person; addressed work first); `fleek orca tasks` lists the project's open tickets. Take only status roles where agents work: intake_review, ready, analysis, design, implementation, review. Skip tickets another holder claims.\n- Read stage in fleek_get_work_context before starting a worker, then use this role table. Do not infer a role from the displayed status name or from the ticket title.\n\n| Status role | Work and delivery |\n| --- | --- |\n| intake_review | One triage agent: duplicates, criteria and readiness; no product code or PR. |\n| ready | Claim, then re-read the live stage and use its row. A claim is not permission to skip analysis or design. |\n| analysis / design | One agent: concept or design proposal with evidence; no product code or PR. |\n| implementation | Builder, other-model review, then deliver the work claim (sections 3-5). |\n| review | One reviewer with a review claim (`fleek orca start KEY --yes --agent OTHER-MODEL` takes it on the delivered branch in its own worktree): checklist, acceptance and evidence; record the verdict, then release that review claim. No builder, implementation, PR or complete_work outcome done loop. Findings go back to the existing implementation work or to the person. |\n| collecting / clarification / acceptance / customer gate | Start no builder; prepare a draft only if requested and wait for the responsible person or customer. Never decide their gate. |\n| done / cancelled | Start nothing. |\n\n- Only implementation uses sections 3-5. `fleek orca tasks` is an overview; `fleek orca pick` works only in a person's terminal and offers only server candidates without an existing work claim. Re-read the live context after claiming: availability is not a status-role decision.\n\n## 3. Claim, worktree, builder\n- `fleek orca start KEY --yes` (the core of `fleek start` for coordinators: Orca required; a new or restarted agent stays in the background, one already running comes to the front) takes or reuses the claim for this machine, creates the Orca worktree named after the key from the project's base branch with the Fleek ticket as reference and starts the agent with a first prompt naming key and claim. You run without a terminal: only `--yes` replaces the person's confirmation, without it the command refuses with CONFIRMATION_REQUIRED. Use it only for tickets your person lets you start; the web link and a person's own terminal keep asking. With the Fleek start command as Orca's agent command (`fleek setup` names it), the agent loads the project's kit. Your own claim is reused, no error. Another holder's claim or NOT_READY: pick another ticket.\n- By hand: fleek_claim_issue (purpose work, runLabel \"orca\" plus the worktree name), then `orca worktree create --name eng-123-slug --reference TICKET-URL --agent claude` from the base branch of fleek_get_project_brief.\n- The builder's task is a pointer: key, claimId, worktree, branch, role builder. Never paste ticket text, criteria or sign-in data; the agent reads them live from Fleek.\n- Models only from providers conventions.aiProviders allows (fleek_get_work_context); a customer's code never goes to another provider.\n- Card: `orca worktree set --worktree id:WORKTREE-ID --workspace-status in-progress`.\n\n## 4. Review by the other model\n- After the builder's worker_done with a pushed commit: start a reviewer with the other model (Claude and Codex swap; only one provider allowed: a fresh session of it) in the same worktree, role reviewer. It claims purpose review, checks criteria and checklist and records its verdict at the ticket.\n- Verdict not ready or a criterion not_met: builder round 2 in the same worktree, then review again. After two rounds ask the person.\n\n## 5. Deliver\n- Clean review: load the guidance (fleek_get_work_context), then fleek_complete_work on the work claim, outcome done, with SHA, branch, PR, checks and result, acceptance per criterion and nextStep. The ticket moves to the project's review state, never Done; check moved and toStateId.\n- A coordination or review claim that only records a status writes fleek_post_progress, not a result. One completion = one fleek_complete_work; no extra summary or analysis comment. Put the recommendation in nextStep; do not add a note about your own draft or publication request.\n- Parts open: outcome partial. Blocked: fleek_release_issue outcome not_doable. Interrupted: outcome paused. A failed dispatch is never done.\n- Card in-review with the existing result; no extra ticket comment.\n\n## 6. People decide\n- Questions only a person can answer go to the ticket (fleek_ask_question); the worker waits, the card gets `--unread`. Gates, acceptance, customer text and Done are decided by a person in Fleek, by an agent only where the project rule allows it and the person approved or gave a standing mandate. Never orca gate-resolve or a terminal yes on their behalf.\n\n## 7. Keep claims alive\n- Renew by expiresAt (about every 20 minutes) while a worker runs or waits; an Orca heartbeat is no renewal. The claim belongs to the machine, so a switched account or a usage-limit pause keeps it: renew and resume. Nobody can continue before expiry: release paused.\n- After a crash read your live claims (fleek_my_work_inbox) and `orca worktree list`, then continue on the same claim (`fleek orca start KEY --yes` restarts an ended agent) or release it. Never claim or complete twice.\n\n## 8. Limits\n- One builder per ticket, one ticket per worktree. Parallel tickets: as many as the person allows, default 2, each with its reviewer; queue the rest. QUOTA_EXCEEDED: the machine's claims are used up - finish or release one first.\n- No Orca code changes, forks or pull requests on Orca. Git effects only as far as the project brief allows.\n"
+    "content": "---\nname: fleek-orca-coordinator\ndescription: \"Coordinate Fleek tickets in Orca: find claimable work, claim it for this machine, start a builder in its own worktree with the project kit, review with the other model, leave gates to people and deliver to review. Use when you coordinate Orca agents for a Fleek project.\"\nmetadata:\n  protocol: fleek-work-protocol/1.13\n---\n\n<!-- Generated by scripts/agent-guidance/generate.mjs from scripts/agent-guidance/protocol.json - do not edit by hand. -->\n\n# Fleek in Orca: coordinator recipe\n\nFleek holds the tickets and every human decision; Orca runs worktrees and agents. Follow the fleek-work skill for every Fleek call. Never change Orca's code or settings.\n\n## 1. Ready check\n- `fleek doctor` in the repo: Fleek MCP current for Claude Code, Codex and every Orca account, repo bound to its Fleek project. Otherwise ask the person to run the setup command from Fleek once per machine (Settings › Connections › Set up this computer: `fleek setup <code>`). Never sign in for them.\n\n## 2. Find work\n- fleek_find_work and fleek_my_work_inbox (claimable: agent-ready or assigned to your person; addressed work first); `fleek orca tasks` lists the project's open tickets. Take only status roles where agents work: intake_review, ready, analysis, design, implementation, review. Skip tickets another holder claims.\n- Read stage in fleek_get_work_context before starting a worker, then use this role table. Do not infer a role from the displayed status name or from the ticket title.\n\n| Status role | Work and delivery |\n| --- | --- |\n| intake_review | One triage agent: duplicates, criteria and readiness; no product code or PR. |\n| ready | Claim, then re-read the live stage and use its row. A claim is not permission to skip analysis or design. |\n| analysis / design | One agent: concept or design proposal with evidence; no product code or PR. |\n| implementation | Builder, other-model review, then deliver the work claim (sections 3-5). |\n| review | One reviewer with a review claim (`fleek orca start KEY --yes --agent OTHER-MODEL` takes it on the delivered branch in its own worktree): checklist, acceptance and evidence; record the verdict, then release that review claim. No builder, implementation, PR or complete_work outcome done loop. Findings go back to the existing implementation work or to the person. |\n| collecting / clarification / acceptance / customer gate | Start no builder; prepare a draft only if requested and wait for the responsible person or customer. Never decide their gate. |\n| done / cancelled | Start nothing. |\n\n- Only implementation uses sections 3-5. `fleek orca tasks` is an overview; `fleek orca pick` works only in a person's terminal and offers only server candidates without an existing work claim. Re-read the live context after claiming: availability is not a status-role decision.\n\n## 3. Claim, worktree, builder\n- `fleek orca start KEY --yes` (the core of `fleek start` for coordinators: Orca required; a new or restarted agent stays in the background, one already running comes to the front) takes or reuses the claim for this machine, creates the Orca worktree named after the key from the project's base branch with the Fleek ticket as reference and starts the agent with a first prompt naming key and claim. You run without a terminal: only `--yes` replaces the person's confirmation, without it the command refuses with CONFIRMATION_REQUIRED. Use it only for tickets your person lets you start; the web link and a person's own terminal keep asking. With the Fleek start command as Orca's agent command (`fleek setup` names it), the agent loads the project's kit. Your own claim is reused, no error. Another holder's claim or NOT_READY: pick another ticket.\n- By hand: fleek_claim_issue (purpose work, runLabel \"orca\" plus the worktree name), then `orca worktree create --name eng-123-slug --reference TICKET-URL --agent claude` from the base branch of fleek_get_project_brief.\n- The builder's task is a pointer: key, claimId, worktree, branch, role builder. Never paste ticket text, criteria or sign-in data; the agent reads them live from Fleek.\n- Models only from providers conventions.aiProviders allows (fleek_get_work_context); a customer's code never goes to another provider.\n- Card: `orca worktree set --worktree id:WORKTREE-ID --workspace-status in-progress`.\n\n## 4. Review by the other model\n- After the builder's worker_done verify its report in two reads before starting a reviewer: the fetched feature branch head equals the reported SHA and the worktree has no uncommitted changes; the test note is on the ticket when the project kit asks for one. A mismatch goes back to the builder, not to the reviewer.\n- Then start a reviewer with the other model (Claude and Codex swap; only one provider allowed: a fresh session of it) in the same worktree, role reviewer. It claims purpose review, checks criteria and checklist and records its verdict at the ticket.\n- Check the review's shape (fleek-work, In Orca, reviewer): verdict on the first line, blockers each with a reproduction, anchors only inside the reviewed range, each finding read or measured, questions for a person at the ticket. A review that misses it goes back to the reviewer once with the exact points.\n- Verdict not ready or a criterion not_met: builder round 2 in the same worktree, then review again. After two rounds ask the person.\n\n## 5. Deliver\n- Clean review: load the guidance (fleek_get_work_context), then fleek_complete_work on the work claim, outcome done, with SHA, branch, PR, checks and result, acceptance per criterion and nextStep. The ticket moves to the project's review state, never Done; check moved and toStateId.\n- A coordination or review claim that only records a status writes fleek_post_progress, not a result. One completion = one fleek_complete_work; no extra summary or analysis comment. Put the recommendation in nextStep; do not add a note about your own draft or publication request.\n- Parts open: outcome partial. Blocked: fleek_release_issue outcome not_doable. Interrupted: outcome paused. A failed dispatch is never done.\n- Card in-review with the existing result; no extra ticket comment.\n\n## 6. People decide\n- Questions only a person can answer go to the ticket (fleek_ask_question); the worker waits, the card gets `--unread`. Gates, acceptance, customer text and Done are decided by a person in Fleek, by an agent only where the project rule allows it and the person approved or gave a standing mandate. Never orca gate-resolve or a terminal yes on their behalf.\n- Answer an orca ask only when a rule or a decision recorded at the ticket settles it, and cite it. Anything else a person decides goes back as \"ask at the ticket\": the asking agent calls fleek_ask_question with the deciding person as addressee and waits. Your own person's answer in chat or terminal is recorded verbatim as an internal ticket comment (\"Answer NAME in chat, DATE: ...\"); pass it on only together with that comment, never a chat answer alone. The comment only records the factual answer of your own person; it decides no gate, customer acceptance or publication and grants or widens no mandate or agent_git_* right - those stay decisions in Fleek.\n\n## 7. Keep claims alive\n- Renew by expiresAt (about every 20 minutes) while a worker runs or waits; an Orca heartbeat is no renewal. The claim belongs to the machine, so a switched account or a usage-limit pause keeps it: renew and resume. Nobody can continue before expiry: release paused.\n- After a crash read your live claims (fleek_my_work_inbox) and `orca worktree list`, then continue on the same claim (`fleek orca start KEY --yes` restarts an ended agent) or release it. Never claim or complete twice.\n\n## 8. Limits\n- One builder per ticket, one ticket per worktree. Parallel tickets: as many as the person allows, default 2, each with its reviewer; queue the rest. QUOTA_EXCEEDED: the machine's claims are used up - finish or release one first.\n- No Orca code changes, forks or pull requests on Orca. Git effects only as far as the project brief allows.\n"
   }
 ];
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@logic-joe/fleek",
-  "version": "0.2.12",
+  "version": "0.2.14",
   "description": "fleek - command line client for the Fleek agent hub (MCP over OAuth, OS keychain, signed role/project kits)",
   "license": "Apache-2.0",
   "type": "module",