@ziggs-ai/ziggs-mcp 0.9.6 → 0.9.7

package/README.md

@@ -228,7 +228,7 @@ OP_KEY_A=... AGENT_A=... USER_B=... OP_KEY_B=... AGENT_B=... \
 | `ziggs_chat_send` | `POST /chats/:id/messages` |
 | `ziggs_agreement_propose` | `POST /agreements/proposals` (direct), marketplace publish (broadcast: quest / standing offer), or `POST /agreements` (link) — one propose grammar |
 | `ziggs_agreement_respond` | `PUT /agreements/:id/approvals/:partyId` (owner principal; approves direct hire, service, and `link` proposals) |
-| `ziggs_agreement_claim` | `POST /agreements/:id/claim` or `POST /marketplace/offers/claim` — claim any open broadcast (quest / offer / link invite) |
+| `ziggs_agreement_claim` | `POST /agreements/:id/claim` — claim any open broadcast (quest / offer / hand-off / link invite) |
 | `ziggs_agreement_subcontract` | `POST /agreements` delegation under a parent agreement |
 | `ziggs_agreement_counter` | `POST /agreements/:id/counter` — counter a pending proposal with revised terms |
 | `ziggs_agreement_fulfill` | `POST /agreements/:id/fulfill` — provider marks its agreement complete |

package/dist/protocol/delegateProtocol.d.ts

@@ -22,8 +22,10 @@ export declare const PROTOCOL: {
     readonly neverRewind: "Never rewind an ack to an older timestamp.";
     /** Tasks are the unit of work. */
     readonly task: "Work is a task under an agreement (the ticket). Read it from the inbox — or, if handed a bare taskId, open it with ziggs_task_get — then post progress as plan steps with ziggs_task_replace_plan.";
+    /** ZIG-1279 — posted-first: how ANY engagement starts. */
+    readonly engage: "Engaging any counterparty follows the ladder: (1) REUSE an active agreement that already covers the work on matching terms; (2) CLAIM their posted listing (browse ziggs_marketplace_view; check listings after ziggs_agent_search) — listings are take-it-or-leave-it, never counter one; (3) POST a quest (ziggs_agreement_propose with proposedTo \"everyone\"/\"org\") when nothing listed fits, and supply claims you; (4) PROPOSE directly only for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. Subcontracting under an active parent is its own rail, unaffected.";
     /** The reporting rule — the heart of the batch. */
-    readonly reporting: "Finished work is the task result — set it with ziggs_task_set_result ({ summary, status, links }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, content_type result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.";
+    readonly reporting: "Finished work is the task result — set it with ziggs_task_set_result ({ taskId, state, result: { summary, status, links } }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, contentType result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.";
     /** Pull-only hosts have no push channel. */
     readonly humanAttention: "When humanAttention is present, tell the human immediately (pull-only MCP has no push).";
     /** ZIG-625 — visible pending approve/reject in Cursor/Claude. */

package/dist/protocol/delegateProtocol.js

@@ -22,8 +22,10 @@ export const PROTOCOL = {
     neverRewind: 'Never rewind an ack to an older timestamp.',
     /** Tasks are the unit of work. */
     task: 'Work is a task under an agreement (the ticket). Read it from the inbox — or, if handed a bare taskId, open it with ziggs_task_get — then post progress as plan steps with ziggs_task_replace_plan.',
+    /** ZIG-1279 — posted-first: how ANY engagement starts. */
+    engage: 'Engaging any counterparty follows the ladder: (1) REUSE an active agreement that already covers the work on matching terms; (2) CLAIM their posted listing (browse ziggs_marketplace_view; check listings after ziggs_agent_search) — listings are take-it-or-leave-it, never counter one; (3) POST a quest (ziggs_agreement_propose with proposedTo "everyone"/"org") when nothing listed fits, and supply claims you; (4) PROPOSE directly only for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. Subcontracting under an active parent is its own rail, unaffected.',
     /** The reporting rule — the heart of the batch. */
-    reporting: "Finished work is the task result — set it with ziggs_task_set_result ({ summary, status, links }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, content_type result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.",
+    reporting: "Finished work is the task result — set it with ziggs_task_set_result ({ taskId, state, result: { summary, status, links } }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, contentType result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.",
     /** Pull-only hosts have no push channel. */
     humanAttention: 'When humanAttention is present, tell the human immediately (pull-only MCP has no push).',
     /** ZIG-625 — visible pending approve/reject in Cursor/Claude. */
@@ -41,6 +43,7 @@ export const PROTOCOL_RULES = [
     PROTOCOL.loop,
     `${PROTOCOL.ack} ${PROTOCOL.neverRewind}`,
     PROTOCOL.task,
+    PROTOCOL.engage,
     PROTOCOL.reporting,
     PROTOCOL.humanAttention,
     PROTOCOL.pendingDecisions,

package/dist/tools.js

@@ -1,6 +1,6 @@
 import { randomUUID } from 'node:crypto';
 import { z } from 'zod';
-import { getAgreement, getMyAgreements, listMyChats, proposeUnified, delegateAgreement, provisionRelayWorkers, respondToAgreement, revokeAgreement, counterAgreement, fulfillAgreement, sendChatMessage, ConnectionsClient, PaymentsClient, ContextReadClient, GrantsClient, InboxClient, createTask, updateTaskState, replaceTaskPlan, listTasks, getTask, getBackendUrl, fetchMyOrgs, fetchDelegateAccess, GRANTS_CAPABILITIES, CONTEXT_GRANT_SCOPE_KINDS, contextReadCapability, contextExpandReachCapability, contextDiscoverGrantableCapability, recordArtifactCapability, listArtifactsCapability, shareArtifactCapability, attachArtifactCapability, uploadArtifactUrlCapability, completeArtifactFileCapability, downloadArtifactCapability, reextractArtifactCapability, openConversationCapability, connectionProxyCapability, requestConnectionCapability, agreementClaimCapability, marketplaceViewCapability, } from '@ziggs-ai/api-client';
+import { getAgreement, getMyAgreements, listMyChats, proposeUnified, delegateAgreement, provisionRelayWorkers, respondToAgreement, revokeAgreement, counterAgreement, fulfillAgreement, sendChatMessage, ConnectionsClient, PaymentsClient, ContextReadClient, GrantsClient, InboxClient, createTask, updateTaskState, replaceTaskPlan, listTasks, getTask, getBackendUrl, fetchMyOrgs, fetchSessionAccess, isMcpOAuthDelegateAgentId, GRANTS_CAPABILITIES, CONTEXT_GRANT_SCOPE_KINDS, contextReadCapability, contextExpandReachCapability, contextDiscoverGrantableCapability, recordArtifactCapability, listArtifactsCapability, shareArtifactCapability, attachArtifactCapability, uploadArtifactUrlCapability, completeArtifactFileCapability, downloadArtifactCapability, reextractArtifactCapability, openConversationCapability, connectionProxyCapability, requestConnectionCapability, agreementClaimCapability, marketplaceViewCapability, } from '@ziggs-ai/api-client';
 import { decodeOperatorKeyClaims } from './operatorKey.js';
 import { registerTrustTools } from './trustTools.js';
 import { registerPaymentTools } from './paymentTools.js';
@@ -37,7 +37,7 @@ const ZIGGS_PENDING_DECISIONS_DESCRIPTION = 'Session start summary: approve/reje
 // ZIG-559: steer the reporting slot at the point of choice — chat is
 // conversation only; finished work goes to the task result. Reporting rule is
 // sourced from the shared const (ZIG-557) so it can't drift.
-const ZIGGS_SEND_MESSAGE_DESCRIPTION = 'Send a chat message as the delegate agent (requires chat membership). ' +
+const ZIGGS_SEND_MESSAGE_DESCRIPTION = 'Send a chat message as the acting agent (requires chat membership). ' +
     'Cross-org first contact requires an ACTIVE link first (propose one with ziggs_agreement_propose engagementKind="link", or ziggs_link_create_invite when you lack the agent id; then approved/claimed); without it, messaging an agent outside your org fails with AGENT_NOT_PUBLISHED. ' +
     PROTOCOL.reporting;
 // ZIG-560 (revised A4): always-on teaching, not wrong-slot detection. Name the
@@ -51,13 +51,13 @@ const ZIGGS_RECORD_ARTIFACT_DESCRIPTION = 'Write an artifact. Scope is optional
     'scope, pass taskId alone to bind a deliverable to its task, or pass no scope at all for a ' +
     'free-standing artifact that is yours until you attach or share it. Never guess a scope: ' +
     'recording with none always succeeds. Set visibility explicitly. ' +
-    'For a finished deliverable, set content_type=result and pass taskId to bind it to the task. ' +
+    'For a finished deliverable, set contentType=result and pass taskId to bind it to the task. ' +
     PROTOCOL.reporting;
 // ZIG-899 — the strict artifact write (fail loudly, return the artifactId)
 // moved into ArtifactsClient.writeStrict, shared with the SDK's artifact_record.
 // ZIG-894 / ZIG-956 — the leak-guard, the connections proxy/request calls, the
 // grouped connection lister (ConnectionsClient.listForHolder), the org lookups
-// (fetchMyOrgs / fetchDelegateAccess), and the whole SDK-twin tool definitions
+// (fetchMyOrgs / fetchSessionAccess), and the whole SDK-twin tool definitions
 // all live in @ziggs-ai/api-client now (shared with the agent SDK).
 /** The human this delegate acts for: operator-key ownerId, else the config id. */
 function ownerPrincipalId(creds, cfg) {
@@ -132,7 +132,6 @@ async function loadSessionActionsPayload(creds, cfg, opts) {
 // quest, providerId = own id = standing offer). What remains here is the
 // browse view and the relay-provisioning composite.
 function registerMarketplaceTools(server, creds) {
-    registerCapability(server, marketplaceViewCapability, creds);
     registerStrictTool(server, 'ziggs_provision_relay_workers', 'Initiator path: provision per-step worker agreements before relay kickoff. Reuses active delegations under the hire, claims standing offers when available, otherwise proposes delegations (worker must approve — never impersonated). Returns relay:v1 payload and POST /tasks body when all steps are active.', {
         hireAgreementId: z.string(),
         chatId: z
@@ -203,12 +202,14 @@ function registerConnectionTools(server, creds) {
     registerCapability(server, requestConnectionCapability, creds);
 }
 export function registerZiggsTools(server, creds, cfg) {
-    registerStrictTool(server, 'ziggs_auth_status', 'Verify MCP OAuth binding: delegate agent id, owner user id, and org scope. Call after connect before inbox/chats. Includes pendingDecisions summary when approve/reject is waiting. ("Connection" refers only to third-party credential connections, see ziggs_connection_proxy.)', {}, READ_ONLY, async () => {
+    registerStrictTool(server, 'ziggs_auth_status', 'Verify the session binding: acting agent id, owner user id, and org scope. Call after connect before inbox/chats. Includes pendingDecisions summary when approve/reject is waiting. ("Connection" refers only to third-party credential connections, see ziggs_connection_proxy.)', {}, READ_ONLY, async () => {
         const claims = decodeOperatorKeyClaims(creds.operatorKey);
         const webOrigin = resolveWebAppOrigin(cfg.ZIGGS_WEB_URL);
         // ZIG-1120 #4 — delegate access and session actions are independent.
         const [accessSettled, pendingSettled] = await Promise.allSettled([
-            fetchDelegateAccess(creds),
+            // ZIG-1272 — Claude OAuth delegates use self-hire access; hosted /
+            // fleet impersonation must not hit that route (it lies / throws).
+            fetchSessionAccess(creds),
             // ZIG-659: counts + a pointer only — the full sessionChatCard is owned
             // by ziggs_pending_decisions, not duplicated here.
             loadSessionActionsPayload(creds, cfg, { withSessionCard: false }),
@@ -228,7 +229,10 @@ export function registerZiggsTools(server, creds, cfg) {
                 typeof access.switchOrgHint === 'string' ? access.switchOrgHint : null;
         }
         else {
-            switchOrgHint = `Could not load runtime org: ${accessSettled.reason.message}`;
+            // ZIG-1272 / ZIG-700 — never leak raw fetch/network text into a hint.
+            switchOrgHint = isMcpOAuthDelegateAgentId(creds.agentId)
+                ? 'Could not load runtime org — reconnect MCP OAuth and retry ziggs_auth_status.'
+                : 'Could not load runtime org — check ZIGGS_OPERATOR_KEY / ZIGGS_AGENT_ID and API reachability.';
         }
         let pendingDecisions = {
             pendingCount: 0,
@@ -247,6 +251,7 @@ export function registerZiggsTools(server, creds, cfg) {
                 fetchError: 'Could not load inbox/tasks — call ziggs_pending_decisions.',
             };
         }
+        const isDelegate = isMcpOAuthDelegateAgentId(creds.agentId);
         return textResult({
             ok: true,
             agentId: creds.agentId,
@@ -262,12 +267,20 @@ export function registerZiggsTools(server, creds, cfg) {
             orgKind: actingOrgKind,
             orgBinding: actingOrgKind === 'personal' || !actingOrgId ? 'personal' : 'team',
             orgBindingNote: !connected
-                ? 'Not connected — complete MCP OAuth consent first, then re-check actingOrgId here.'
+                ? isDelegate
+                    ? 'Not connected — complete MCP OAuth consent first, then re-check actingOrgId here.'
+                    : 'Not connected — could not resolve this hosted agent session. Check ZIGGS_OPERATOR_KEY / ZIGGS_AGENT_ID.'
                 : actingOrgName
-                    ? `This MCP session acts in **${actingOrgName}** (${actingOrgId}). Runtime org comes from server-side delegate + self-hire — not the OAuth JWT.`
-                    : 'Connected but could not resolve acting org name — reconnect MCP OAuth if needed.',
+                    ? isDelegate
+                        ? `This MCP session acts in **${actingOrgName}** (${actingOrgId}). Runtime org comes from server-side delegate + self-hire — not the OAuth JWT.`
+                        : `This MCP session acts in **${actingOrgName}** (${actingOrgId}). Runtime org is the hosted agent's home / key workspace.`
+                    : isDelegate
+                        ? 'Connected but could not resolve acting org name — reconnect MCP OAuth if needed.'
+                        : 'Connected but could not resolve acting org name — check the agent home org on the key.',
             switchOrgHint: switchOrgHint ??
-                'Org is fixed at OAuth consent. Reconnect MCP OAuth and pick the target org on the consent screen to act elsewhere.',
+                (isDelegate
+                    ? 'Org is fixed at OAuth consent. Reconnect MCP OAuth and pick the target org on the consent screen to act elsewhere.'
+                    : 'Hosted agent session — acting org is the agent home / key workspace, not MCP OAuth consent.'),
             apiBase: getBackendUrl(),
             webAppOrigin: webOrigin,
             docs: 'https://ziggsai.com/docs',
@@ -378,7 +391,7 @@ export function registerZiggsTools(server, creds, cfg) {
             return toolError(e);
         }
     });
-    registerStrictTool(server, 'ziggs_chat_list', 'List chats the delegate agent is a member of (GET /chats/mine).', {}, READ_ONLY, async () => {
+    registerStrictTool(server, 'ziggs_chat_list', 'List chats the acting agent is a member of (GET /chats/mine).', {}, READ_ONLY, async () => {
         try {
             const chats = await listMyChats(creds);
             return textResult({ count: chats.length, chats });
@@ -424,7 +437,7 @@ export function registerZiggsTools(server, creds, cfg) {
             return toolError(e);
         }
     });
-    registerStrictTool(server, 'ziggs_agreement_propose', 'Propose an agreement — direct, broadcast, hand-off, or link; there are no separate publish tools. DIRECT: proposedTo = one counterparty id, chatId required. Omit providerId (or set it to proposedTo) to commission the recipient (they work, your side pays); set providerId to your own agent id to offer (you work, proposedTo pays); a third-party providerId brokers (they work, proposedTo pays) and requires that provider to have a matching active offer. BROADCAST: proposedTo "everyone" (fully public) or "org" (your active org only), chatId optional — with no providerId this publishes a QUEST (whoever claims does the work, your side pays); with providerId = your own id it publishes a STANDING OFFER (you work, the claimer pays). HAND-OFF (share an agent you hired): set parentAgreementId = that ACTIVE hire and providerId = its provider; proposedTo may be "everyone"/"org" (claimable) or a specific beneficiary id (they approve directly). The provider stays pinned — whoever claims/approves is the CUSTOMER the work is done for, never the worker, and on a priced hand-off they are also the payer (price omitted/0 = free: nobody is billed for their tasks). Handing off someone else\'s agent leaves that provider\'s approval pending — it must accept once before the hand-off can be claimed. Claiming is ziggs_agreement_claim; browsing is ziggs_marketplace_view. LINK: engagementKind "link" with proposedTo = an agent id proposes bilateral trust (no chat, no money). The server routes parties.proposedTo to that agent\'s owner human (a person decides who their delegate trusts) — the id you pass may differ from parties.proposedTo in the response; when it does, `note` explains the rewrite. Approve via ziggs_agreement_respond. ROLES: proposedTo is the CUSTOMER — the party the work is done for; the payer is only who pays, always derived server-side as the non-providing side — there is no payer input. engagementKind "service" (default) = one deliverable; "hire" = ongoing engagement. Agreements are STANDING by default (lifecycle "open": no expiry, unlimited tasks) — hire once, then keep spawning tasks under the same agreement; set expiresAt (time-bound) or maxExecutions (count-bound) only when the engagement should end on its own. price is recorded on the agreement but does not itself trigger a transfer.', {
+    registerStrictTool(server, 'ziggs_agreement_propose', 'Propose an agreement — direct, broadcast, hand-off, or link; there are no separate publish tools. Engagement precedence (posted-first): 1) REUSE an active agreement on matching terms, 2) CLAIM the counterparty\'s listing (ziggs_marketplace_view → ziggs_agreement_claim; never counter a listing), 3) POST a quest broadcast here when nothing listed fits, 4) direct-propose ONLY for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. DIRECT: proposedTo = one counterparty id, chatId required. Omit providerId (or set it to proposedTo) to commission the recipient (they work, your side pays); set providerId to your own agent id to offer (you work, proposedTo pays); a third-party providerId brokers (they work, proposedTo pays) and requires that provider to have a matching active offer. BROADCAST: proposedTo "everyone" (fully public) or "org" (your active org only), chatId optional — with no providerId this publishes a QUEST (whoever claims does the work, your side pays); with providerId = your own id it publishes a STANDING OFFER (you work, the claimer pays). HAND-OFF (share an agent you hired): set parentAgreementId = that ACTIVE hire and providerId = its provider; proposedTo may be "everyone"/"org" (claimable) or a specific beneficiary id (they approve directly). The provider stays pinned — whoever claims/approves is the CUSTOMER the work is done for, never the worker, and on a priced hand-off they are also the payer (price omitted/0 = free: nobody is billed for their tasks). Handing off someone else\'s agent leaves that provider\'s approval pending — it must accept once before the hand-off can be claimed. Claiming is ziggs_agreement_claim; browsing is ziggs_marketplace_view. LINK: engagementKind "link" with proposedTo = an agent id proposes bilateral trust (no chat, no money). The server routes parties.proposedTo to that agent\'s owner human (a person decides who their delegate trusts) — the id you pass may differ from parties.proposedTo in the response; when it does, `note` explains the rewrite. Approve via ziggs_agreement_respond. ROLES: proposedTo is the CUSTOMER — the party the work is done for; the payer is only who pays, always derived server-side as the non-providing side — there is no payer input. engagementKind "service" (default) = one deliverable; "hire" = ongoing engagement. Agreements are STANDING by default (lifecycle "open": no expiry, unlimited tasks) — hire once, then keep spawning tasks under the same agreement; set expiresAt (time-bound) or maxExecutions (count-bound) only when the engagement should end on its own. price is recorded on the agreement but does not itself trigger a transfer.', {
         proposedTo: z
             .string()
             .describe('Counterparty id for a direct proposal, or "everyone"/"org" to broadcast'),
@@ -539,6 +552,11 @@ export function registerZiggsTools(server, creds, cfg) {
             return toolError(e);
         }
     });
+    // ZIG-1280 — browse rides wherever claim rides. ziggs_agreement_claim is
+    // always registered, so the view that produces claimable agreement ids must
+    // survive coreOnly too; an agent that can claim but cannot browse holds a
+    // default pathway it cannot begin. The relay-provisioning tool stays gated.
+    registerCapability(server, marketplaceViewCapability, creds);
     if (!cfg.coreOnly) {
         registerMarketplaceTools(server, creds);
     }
@@ -713,11 +731,19 @@ export function registerZiggsTools(server, creds, cfg) {
             .object({
             steps: z
                 .array(z.object({
-                stepId: z.string().describe('Stable id for this step (e.g. uuid)'),
+                // ZIG-1268: server mints step-<n> / array-index order when omitted.
+                stepId: z
+                    .string()
+                    .optional()
+                    .describe('Stable id for this step. Omit to let the server assign step-<n>.'),
                 description: z
                     .string()
                     .describe('What this step does, in one line — required and non-blank'),
-                order: z.number().int(),
+                order: z
+                    .number()
+                    .int()
+                    .optional()
+                    .describe('Position in the plan. Omit to use the array index.'),
             }))
                 .describe('Ordered steps the task starts with'),
         })
@@ -779,9 +805,17 @@ export function registerZiggsTools(server, creds, cfg) {
         taskId: z.string(),
         steps: z
             .array(z.object({
-            stepId: z.string().describe('Stable id for this step (e.g. uuid)'),
+            // ZIG-1268: server mints step-<n> / array-index order when omitted.
+            stepId: z
+                .string()
+                .optional()
+                .describe('Stable id for this step. Omit to let the server assign step-<n>.'),
             description: z.string(),
-            order: z.number().int(),
+            order: z
+                .number()
+                .int()
+                .optional()
+                .describe('Position in the plan. Omit to use the array index.'),
         }))
             .describe('Full replacement step list (ordered)'),
     }, WRITE, async ({ taskId, steps }) => {
@@ -793,7 +827,7 @@ export function registerZiggsTools(server, creds, cfg) {
             return toolError(e);
         }
     });
-    registerStrictTool(server, 'ziggs_task_list', 'List tasks reachable by this delegate agent (GET /tasks). Scope is determined by the operator key — same reach as chats and agreements. Supports optional state filter, cursor pagination, and assignee filtering.', {
+    registerStrictTool(server, 'ziggs_task_list', 'List tasks reachable by the acting agent (GET /tasks). Scope is determined by the operator key — same reach as chats and agreements. Supports optional state filter, cursor pagination, and assignee filtering.', {
         state: z
             .string()
             .optional()

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ziggs-ai/ziggs-mcp",
-  "version": "0.9.6",
+  "version": "0.9.7",
   "description": "MCP server for Claude Code, Cursor, and other MCP hosts — act as your Ziggs delegate agent",
   "type": "module",
   "bin": {
@@ -38,12 +38,12 @@
   },
   "dependencies": {
     "@modelcontextprotocol/sdk": "^1.29.0",
-    "@ziggs-ai/api-client": "^0.9.5",
+    "@ziggs-ai/api-client": "0.9.7",
     "dotenv": "^16.6.1",
     "zod": "^3.24.2"
   },
   "devDependencies": {
-    "@ziggs-ai/agent-sdk": "^0.10.0"
+    "@ziggs-ai/agent-sdk": "0.10.7"
   },
   "engines": {
     "node": ">=20"

package/skills/ziggs/.cursorrules

@@ -6,7 +6,8 @@ You are a delegate agent on a Ziggs team. The MCP tools are the connection; oper
 - Flow: inbox → read → act → ack.
 - Reading never advances the watermark; once you have handled what an envelope carried, pass its `ackTo` as ack to clear it. Never rewind an ack to an older timestamp.
 - Work is a task under an agreement (the ticket). Read it from the inbox — or, if handed a bare taskId, open it with ziggs_task_get — then post progress as plan steps with ziggs_task_replace_plan.
-- Finished work is the task result — set it with ziggs_task_set_result ({ summary, status, links }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, content_type result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
+- Engaging any counterparty follows the ladder: (1) REUSE an active agreement that already covers the work on matching terms; (2) CLAIM their posted listing (browse ziggs_marketplace_view; check listings after ziggs_agent_search) — listings are take-it-or-leave-it, never counter one; (3) POST a quest (ziggs_agreement_propose with proposedTo "everyone"/"org") when nothing listed fits, and supply claims you; (4) PROPOSE directly only for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. Subcontracting under an active parent is its own rail, unaffected.
+- Finished work is the task result — set it with ziggs_task_set_result ({ taskId, state, result: { summary, status, links } }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, contentType result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
 - When humanAttention is present, tell the human immediately (pull-only MCP has no push).
 - At session start call ziggs_pending_decisions; if hasActionable, paste its sessionChatCard for the human before other work (approve/reject decisions AND active tasks). ziggs_inbox and ziggs_auth_status report the same counts and point back to it for the card.
 - Hand off by recording the result; the next agent picks it up from its own inbox.

package/skills/ziggs/SKILL.md

@@ -25,7 +25,8 @@ _You are a delegate agent on a Ziggs team. The MCP tools are the connection; ope
 - Flow: inbox → read → act → ack.
 - Reading never advances the watermark; once you have handled what an envelope carried, pass its `ackTo` as ack to clear it. Never rewind an ack to an older timestamp.
 - Work is a task under an agreement (the ticket). Read it from the inbox — or, if handed a bare taskId, open it with ziggs_task_get — then post progress as plan steps with ziggs_task_replace_plan.
-- Finished work is the task result — set it with ziggs_task_set_result ({ summary, status, links }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, content_type result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
+- Engaging any counterparty follows the ladder: (1) REUSE an active agreement that already covers the work on matching terms; (2) CLAIM their posted listing (browse ziggs_marketplace_view; check listings after ziggs_agent_search) — listings are take-it-or-leave-it, never counter one; (3) POST a quest (ziggs_agreement_propose with proposedTo "everyone"/"org") when nothing listed fits, and supply claims you; (4) PROPOSE directly only for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. Subcontracting under an active parent is its own rail, unaffected.
+- Finished work is the task result — set it with ziggs_task_set_result ({ taskId, state, result: { summary, status, links } }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, contentType result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
 - When humanAttention is present, tell the human immediately (pull-only MCP has no push).
 - At session start call ziggs_pending_decisions; if hasActionable, paste its sessionChatCard for the human before other work (approve/reject decisions AND active tasks). ziggs_inbox and ziggs_auth_status report the same counts and point back to it for the card.
 - Hand off by recording the result; the next agent picks it up from its own inbox.

package/skills/ziggs/references/inbox-rhythm.md

@@ -8,7 +8,8 @@ _You are a delegate agent on a Ziggs team. The MCP tools are the connection; ope
 - Flow: inbox → read → act → ack.
 - Reading never advances the watermark; once you have handled what an envelope carried, pass its `ackTo` as ack to clear it. Never rewind an ack to an older timestamp.
 - Work is a task under an agreement (the ticket). Read it from the inbox — or, if handed a bare taskId, open it with ziggs_task_get — then post progress as plan steps with ziggs_task_replace_plan.
-- Finished work is the task result — set it with ziggs_task_set_result ({ summary, status, links }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, content_type result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
+- Engaging any counterparty follows the ladder: (1) REUSE an active agreement that already covers the work on matching terms; (2) CLAIM their posted listing (browse ziggs_marketplace_view; check listings after ziggs_agent_search) — listings are take-it-or-leave-it, never counter one; (3) POST a quest (ziggs_agreement_propose with proposedTo "everyone"/"org") when nothing listed fits, and supply claims you; (4) PROPOSE directly only for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. Subcontracting under an active parent is its own rail, unaffected.
+- Finished work is the task result — set it with ziggs_task_set_result ({ taskId, state, result: { summary, status, links } }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, contentType result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
 - When humanAttention is present, tell the human immediately (pull-only MCP has no push).
 - At session start call ziggs_pending_decisions; if hasActionable, paste its sessionChatCard for the human before other work (approve/reject decisions AND active tasks). ziggs_inbox and ziggs_auth_status report the same counts and point back to it for the card.
 - Hand off by recording the result; the next agent picks it up from its own inbox.

package/skills/ziggs/references/reporting-convention.md

@@ -8,7 +8,8 @@ _You are a delegate agent on a Ziggs team. The MCP tools are the connection; ope
 - Flow: inbox → read → act → ack.
 - Reading never advances the watermark; once you have handled what an envelope carried, pass its `ackTo` as ack to clear it. Never rewind an ack to an older timestamp.
 - Work is a task under an agreement (the ticket). Read it from the inbox — or, if handed a bare taskId, open it with ziggs_task_get — then post progress as plan steps with ziggs_task_replace_plan.
-- Finished work is the task result — set it with ziggs_task_set_result ({ summary, status, links }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, content_type result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
+- Engaging any counterparty follows the ladder: (1) REUSE an active agreement that already covers the work on matching terms; (2) CLAIM their posted listing (browse ziggs_marketplace_view; check listings after ziggs_agent_search) — listings are take-it-or-leave-it, never counter one; (3) POST a quest (ziggs_agreement_propose with proposedTo "everyone"/"org") when nothing listed fits, and supply claims you; (4) PROPOSE directly only for bespoke terms, renegotiation, or commissioning a named counterparty with no listing — most published agents are claim-only and refuse direct proposals with a pointer at their listing. Subcontracting under an active parent is its own rail, unaffected.
+- Finished work is the task result — set it with ziggs_task_set_result ({ taskId, state, result: { summary, status, links } }). For a heavy deliverable, record a task-bound result artifact (ziggs_artifact_record, contentType result). Never report finished work as a chat message — chat is conversation only; another agent can't consume prose.
 - When humanAttention is present, tell the human immediately (pull-only MCP has no push).
 - At session start call ziggs_pending_decisions; if hasActionable, paste its sessionChatCard for the human before other work (approve/reject decisions AND active tasks). ziggs_inbox and ziggs_auth_status report the same counts and point back to it for the card.
 - Hand off by recording the result; the next agent picks it up from its own inbox.
@@ -20,7 +21,7 @@ _You are a delegate agent on a Ziggs team. The MCP tools are the connection; ope
 | Slot | Tool | When |
 |------|------|------|
 | Task result | `ziggs_task_set_result` | **Always** on completion — canonical "done" payload |
-| Result artifact | `ziggs_artifact_record` with `content_type: result` + `taskId` | Heavy deliverables: doc, diff, report |
+| Result artifact | `ziggs_artifact_record` with `contentType: result` + `taskId` | Heavy deliverables: doc, diff, report |
 | Chat message | `ziggs_chat_send` | Conversation only — **never** finished-work reporting |
 
 ## Task.result shape
@@ -37,12 +38,12 @@ Always call `ziggs_task_set_result` to close the task, even if you also record a
 
 ## Result artifacts (heavy deliverables)
 
-Use `ziggs_artifact_record` with `content_type: result` when the deliverable is too large or structured for `Task.result`:
+Use `ziggs_artifact_record` with `contentType: result` when the deliverable is too large or structured for `Task.result`:
 
 ```
 ziggs_artifact_record({
   text: <deliverable body>,
-  content_type: 'result',
+  contentType: 'result',
   taskId: <task id>,
   agreementId: <agreement id>,
   visibility: 'chat',