Python
// Synthetic, executable tool payloads. The gallery never classifies tasks from prose.
export const launchProposal = {
title: "Give every customer a launch plan",
summary:
"Turn a won deal into an owned, dated plan—and show what changes when a milestone slips.",
targets: ["acme/billing-service · local workspace"],
external_actions: {
status: "none",
actions: [],
exclusions: ["Pushes, deployments, and outbound messages"],
explanation: "This rehearsal is limited to local changes and verification.",
},
activities: [
{ id: "build", title: "Build", summary: "Implement the customer journey" },
{
id: "verify",
title: "Verify",
summary: "Independently check the results",
},
{ id: "accept", title: "Accept", summary: "Review the complete outcome" },
],
workstreams: [
{
id: "foundation",
title: "Plan, schedule & dependencies",
summary: "Shared data and scheduling foundation",
},
{
id: "experience",
title: "Workspace, timeline & portfolio",
summary: "Independent customer-facing surfaces",
},
{
id: "followthrough",
title: "Templates & reminders",
summary: "Reusable plans and local follow-through",
},
{
id: "quality",
title: "Verification & acceptance",
summary: "Independent evidence and final judgment",
},
],
items: [
{
key: "plan",
title: "Launch plan data & API",
activity: "build",
workstream: "foundation",
depends_on: [],
verifies: [],
criteria:
"A won deal creates an owned plan; existing accounts remain intact.",
},
{
key: "graph",
title: "Dependencies & impact preview",
activity: "build",
workstream: "foundation",
depends_on: [],
verifies: [],
criteria:
"A three-day slip previews affected successors before applying changes; dependency cycles are rejected.",
},
{
key: "workspace",
title: "Launch workspace",
activity: "build",
workstream: "experience",
depends_on: [],
verifies: [],
criteria: "Owners can manage tasks without bypassing required work.",
},
{
key: "timeline",
title: "Interactive timeline",
activity: "build",
workstream: "experience",
depends_on: [],
verifies: [],
criteria: "Dates match the workspace after a schedule change.",
},
{
key: "portfolio",
title: "Portfolio attention",
activity: "build",
workstream: "experience",
depends_on: [],
verifies: [],
criteria:
"Teams can find overdue launches with correct access boundaries.",
},
{
key: "templates",
title: "Versioned launch templates",
activity: "build",
workstream: "followthrough",
depends_on: [],
verifies: [],
criteria:
"New launches use the chosen version; existing plans are unchanged.",
},
{
key: "reminders",
title: "Local reminders",
activity: "build",
workstream: "followthrough",
depends_on: [],
verifies: [],
criteria: "Due-date changes update in-app reminders without duplicates.",
},
{
key: "api-check",
title: "API & permissions",
activity: "verify",
workstream: "quality",
depends_on: ["plan", "graph"],
verifies: ["plan", "graph"],
criteria:
"Tests cover tenant boundaries, roles, completion rules, and scheduling.",
},
{
key: "browser-check",
title: "Browser journey & evidence",
activity: "verify",
workstream: "quality",
depends_on: ["workspace", "timeline", "portfolio"],
verifies: ["workspace", "timeline", "portfolio"],
criteria:
"The full journey passes with screenshots attached to board items.",
},
{
key: "upgrade-check",
title: "Upgrade & regression",
activity: "verify",
workstream: "quality",
depends_on: ["templates", "reminders"],
verifies: ["templates", "reminders"],
criteria: "Existing data survives and the previous journey still works.",
},
{
key: "accept",
title: "Accept the complete journey",
activity: "accept",
workstream: "quality",
depends_on: ["api-check", "browser-check", "upgrade-check"],
verifies: [],
criteria:
"Win a deal, adjust a milestone, verify affected views, and complete the plan.",
},
],
final_acceptance: { item_key: "accept", owner: "lead" },
};
export const securityProposal = {
title: "Close the staging exposure gaps",
summary:
"Assess the approved staging targets, prepare fixes, and independently recheck the findings.",
targets: ["acme/billing-service", "sandbox-1 · approved staging environment"],
external_actions: {
status: "planned",
actions: [
"Run authenticated checks against sandbox-1",
"Open remediation pull requests in acme/billing-service",
],
exclusions: [
"Production scans or infrastructure changes",
"Merging remediation pull requests",
],
explanation:
"Only the staging targets named in the request are in scope. Active scans contact external systems.",
},
activities: [
{ id: "assess", title: "Assess", summary: "Establish exposure" },
{ id: "remediate", title: "Remediate", summary: "Prepare fixes" },
{ id: "validate", title: "Validate", summary: "Independently recheck" },
],
workstreams: [
{
id: "application",
title: "Application security",
summary: "Repository and dependency findings",
},
{
id: "cloud",
title: "Staging infrastructure",
summary: "Exposure and configuration checks",
},
{
id: "review",
title: "Remediation & validation",
summary: "Evidence-backed closure",
},
],
items: [
{
key: "code",
title: "Assess repository findings",
activity: "assess",
workstream: "application",
depends_on: [],
verifies: [],
criteria:
"Findings identify affected code and include reproducible evidence.",
},
{
key: "infra",
title: "Assess staging exposure",
activity: "assess",
workstream: "cloud",
depends_on: [],
verifies: [],
criteria:
"Checks stay within the approved staging targets and report coverage gaps.",
},
{
key: "fix",
title: "Prepare remediation changes",
activity: "remediate",
workstream: "review",
depends_on: ["code", "infra"],
verifies: [],
criteria:
"Each proposed fix references its finding and includes regression coverage.",
},
{
key: "validate",
title: "Accept independently verified findings",
activity: "validate",
workstream: "review",
depends_on: ["fix"],
verifies: ["fix"],
criteria:
"Reported vulnerabilities no longer reproduce; unresolved risks remain explicitly documented.",
},
],
final_acceptance: { item_key: "validate", owner: "assigned_worker" },
};
export const marketingProposal = {
title: "Prepare the customer launch campaign",
summary:
"Research the audience, create campaign drafts, and review claims before publication is considered.",
targets: ["Acme · customer launch campaign"],
external_actions: {
status: "undetermined",
actions: [],
exclusions: [
"Publishing before the user chooses channels and approves the copy",
],
explanation:
"Publication channels and recipients have not been selected. Clarification is needed before any publishing or sending.",
},
activities: [
{ id: "research", title: "Research", summary: "Establish the audience" },
{ id: "create", title: "Create", summary: "Draft assets" },
{ id: "review", title: "Review", summary: "Verify claims and tone" },
],
workstreams: [
{
id: "campaign",
title: "Campaign preparation",
summary: "Audience, drafts, and publication-ready review",
},
],
items: [
{
key: "research",
title: "Audience and message brief",
activity: "research",
workstream: "campaign",
depends_on: [],
verifies: [],
criteria:
"The brief identifies the audience and cites sources for factual claims.",
},
{
key: "draft",
title: "Campaign drafts",
activity: "create",
workstream: "campaign",
depends_on: ["research"],
verifies: [],
criteria:
"Drafts match the approved brief and make no unsupported claims.",
},
{
key: "review",
title: "Review campaign assets",
activity: "review",
workstream: "campaign",
depends_on: ["draft"],
verifies: ["draft"],
criteria:
"Claims, tone, and links are checked; publication requires a separate decision.",
},
],
final_acceptance: { item_key: "review", owner: "lead" },
};
export const launchTeam = {
title: "The customer launch team",
summary:
"Seven builders own separate workstreams. Three verifiers independently check their results.",
groups: [
{
id: "builders",
title: "Builders",
summary: "One owner per implementation workstream",
},
{
id: "verifiers",
title: "Verifiers",
summary: "Independent checks and board evidence",
},
],
members: launchProposal.items.slice(0, 10).map((task, i) => ({
persona: i < 7 ? "swe-worker" : "test-worker",
name: `${i < 7 ? "sam" : "maya"}-${i + 1}`,
reason: task.title,
group: i < 7 ? "builders" : "verifiers",
item_ids: [i + 1],
connectors: [],
resolved_model: "anthropic:claude-sonnet-4-8",
})),
enable_chat: true,
planned_items: launchProposal.items.map((task, i) => ({
id: i + 1,
title: task.title,
final_acceptance: {
id: 11,
title: "Accept the complete journey",
owner: "lead",
},
})),
lead_mode: "interactive",
offer: {},
other_connected: [],
runnable_models: [
{ id: "anthropic:claude-sonnet-4-8", label: "Claude Sonnet 4.8" },
],
lead_model: "anthropic:claude-sonnet-4-8",
model_options: {},
};