@pixelbyte-software/pixcode 1.54.7 → 1.54.9

package/server/modules/orchestration/workflows/workflow-templates.ts

@@ -1,272 +1,272 @@
-export const PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL = 'pixcode.workflow-template.v1' as const;
-
-export interface WorkflowTemplateAgentSlot {
-  id: string;
-  role: string;
-  label: string;
-  instruction: string;
-}
-
-export interface WorkflowTemplate {
-  id: string;
-  protocol: typeof PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL;
-  version: 1;
-  workflowId: string;
-  name: string;
-  description: string;
-  inputPlaceholder: string;
-  agentSlots: WorkflowTemplateAgentSlot[];
-  acceptanceCriteria: string[];
-  defaultSettings?: {
-    maxParallelAgents?: number;
-    maxRepairCycles?: number;
-  };
-}
-
-type AgentRecord = Record<string, unknown>;
-
-export const builtInWorkflowTemplates: WorkflowTemplate[] = [
-  {
-    id: 'bug_fix_team',
-    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
-    version: 1,
-    workflowId: 'agent_team',
-    name: 'Bug fix team',
-    description: 'Reproduce, fix, and verify a focused bug with an implementation and review loop.',
-    inputPlaceholder: 'Describe the bug, expected behavior, current behavior, and any reproduction steps.',
-    agentSlots: [
-      {
-        id: 'reproducer',
-        role: 'implementation',
-        label: 'Reproducer',
-        instruction: 'Reproduce the bug from the user report, identify the failing path, and keep the scope narrow.',
-      },
-      {
-        id: 'fixer',
-        role: 'implementation',
-        label: 'Fixer',
-        instruction: 'Implement the smallest durable fix and report changed files and verification commands.',
-      },
-      {
-        id: 'reviewer',
-        role: 'review',
-        label: 'Reviewer',
-        instruction: 'Review the fix for regressions, missing validation, and whether the original bug is covered.',
-      },
-    ],
-    acceptanceCriteria: [
-      'The bug has a concrete reproduction or a clearly documented blocker.',
-      'The fix is limited to the failing behavior.',
-      'Verification commands and remaining risks are reported.',
-    ],
-    defaultSettings: {
-      maxParallelAgents: 2,
-      maxRepairCycles: 1,
-    },
-  },
-  {
-    id: 'pr_review_team',
-    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
-    version: 1,
-    workflowId: 'multi_model_review',
-    name: 'PR review team',
-    description: 'Run independent reviewers and aggregate actionable findings for a pull request or diff.',
-    inputPlaceholder: 'Paste the PR link, branch, or diff summary to review.',
-    agentSlots: [
-      {
-        id: 'correctness',
-        role: 'review',
-        label: 'Correctness reviewer',
-        instruction: 'Prioritize correctness bugs, regressions, missing tests, and broken edge cases.',
-      },
-      {
-        id: 'security',
-        role: 'review',
-        label: 'Security reviewer',
-        instruction: 'Prioritize security, secret handling, permission, and data exposure risks.',
-      },
-      {
-        id: 'reporter',
-        role: 'report',
-        label: 'Report aggregator',
-        instruction: 'Aggregate findings by severity with file references and concrete fix suggestions.',
-      },
-    ],
-    acceptanceCriteria: [
-      'Findings are ordered by severity and avoid speculative noise.',
-      'Each issue includes the concrete impact and suggested fix.',
-      'The final report clearly says when no blocking issue is found.',
-    ],
-    defaultSettings: {
-      maxParallelAgents: 3,
-    },
-  },
-  {
-    id: 'frontend_polish',
-    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
-    version: 1,
-    workflowId: 'agent_team',
-    name: 'Frontend polish',
-    description: 'Improve a UI workflow with implementation, responsive checks, and UX review.',
-    inputPlaceholder: 'Describe the screen or workflow that needs frontend polish.',
-    agentSlots: [
-      {
-        id: 'ui_implementation',
-        role: 'frontend',
-        label: 'UI implementer',
-        instruction: 'Implement the requested UI change using existing design conventions and responsive constraints.',
-      },
-      {
-        id: 'ux_review',
-        role: 'review',
-        label: 'UX reviewer',
-        instruction: 'Review layout, text overflow, responsive behavior, accessibility, and interaction states.',
-      },
-      {
-        id: 'verification',
-        role: 'review',
-        label: 'Verification reviewer',
-        instruction: 'Verify the change with available typecheck, lint, build, smoke, or manual UI evidence.',
-      },
-    ],
-    acceptanceCriteria: [
-      'The UI follows existing components and visual density.',
-      'Text and controls do not overlap on mobile or desktop.',
-      'Verification evidence is included in the final summary.',
-    ],
-    defaultSettings: {
-      maxParallelAgents: 2,
-      maxRepairCycles: 1,
-    },
-  },
-  {
-    id: 'release_manager',
-    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
-    version: 1,
-    workflowId: 'sequential_handoff',
-    name: 'Release manager',
-    description: 'Prepare a release checklist, run verification, and summarize publish readiness.',
-    inputPlaceholder: 'Describe the version, release scope, and target channels.',
-    agentSlots: [
-      {
-        id: 'release_plan',
-        role: 'implementation',
-        label: 'Release planner',
-        instruction: 'Create a release checklist, verify version surfaces, and identify publish blockers.',
-      },
-      {
-        id: 'release_verification',
-        role: 'review',
-        label: 'Release verifier',
-        instruction: 'Run or inspect release verification commands and report the exact publish state.',
-      },
-      {
-        id: 'release_report',
-        role: 'report',
-        label: 'Release reporter',
-        instruction: 'Summarize what was released, what was verified, and any blocked channels.',
-      },
-    ],
-    acceptanceCriteria: [
-      'Version surfaces are aligned.',
-      'Build/package verification is reported.',
-      'Remote artifact and publish state are explicitly stated.',
-    ],
-    defaultSettings: {
-      maxParallelAgents: 1,
-    },
-  },
-  {
-    id: 'dependency_audit',
-    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
-    version: 1,
-    workflowId: 'agent_team',
-    name: 'Dependency audit',
-    description: 'Inspect dependency, runtime, and package risks without making broad upgrades by default.',
-    inputPlaceholder: 'Describe the dependency or runtime area to audit.',
-    agentSlots: [
-      {
-        id: 'audit',
-        role: 'implementation',
-        label: 'Audit analyst',
-        instruction: 'Inspect dependency and runtime risk using existing lockfiles and configured package managers.',
-      },
-      {
-        id: 'risk_review',
-        role: 'review',
-        label: 'Risk reviewer',
-        instruction: 'Validate the findings, call out false positives, and avoid broad upgrade churn unless requested.',
-      },
-      {
-        id: 'report',
-        role: 'report',
-        label: 'Audit reporter',
-        instruction: 'Produce a prioritized remediation plan with commands, risks, and no-secret output.',
-      },
-    ],
-    acceptanceCriteria: [
-      'Findings are tied to actual dependency evidence.',
-      'No unrelated dependency upgrades are proposed as automatic work.',
-      'Security and runtime risks are separated from maintenance suggestions.',
-    ],
-    defaultSettings: {
-      maxParallelAgents: 2,
-    },
-  },
-];
-
-export function getWorkflowTemplate(templateId: string): WorkflowTemplate | undefined {
-  return builtInWorkflowTemplates.find((template) => template.id === templateId);
-}
-
-function applyTemplateSlots(agents: AgentRecord[] | undefined, template: WorkflowTemplate): AgentRecord[] | undefined {
-  if (!Array.isArray(agents) || agents.length === 0) return agents;
-
-  let enabledSlotIndex = 0;
-  return agents.map((agent) => {
-    if (agent.enabled === false) return agent;
-    const slot = template.agentSlots[enabledSlotIndex];
-    enabledSlotIndex += 1;
-    if (!slot) return agent;
-    return {
-      ...agent,
-      role: slot.role,
-      instruction: typeof agent.instruction === 'string' && agent.instruction.trim()
-        ? agent.instruction
-        : slot.instruction,
-      templateSlotId: slot.id,
-      templateSlotLabel: slot.label,
-    };
-  });
-}
-
-export function applyWorkflowTemplateToMetadata(
-  template: WorkflowTemplate,
-  metadata?: Record<string, unknown>,
-): Record<string, unknown> {
-  const settings = metadata?.settings && typeof metadata.settings === 'object'
-    ? metadata.settings as Record<string, unknown>
-    : {};
-
-  return {
-    ...metadata,
-    workflowTemplate: {
-      protocol: template.protocol,
-      id: template.id,
-      version: template.version,
-      name: template.name,
-      workflowId: template.workflowId,
-      acceptanceCriteria: template.acceptanceCriteria,
-      agentSlots: template.agentSlots.map((slot) => ({
-        id: slot.id,
-        role: slot.role,
-        label: slot.label,
-      })),
-    },
-    agents: applyTemplateSlots(metadata?.agents as AgentRecord[] | undefined, template),
-    settings: {
-      ...settings,
-      ...template.defaultSettings,
-    },
-  };
-}
+export const PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL = 'pixcode.workflow-template.v1' as const;
+
+export interface WorkflowTemplateAgentSlot {
+  id: string;
+  role: string;
+  label: string;
+  instruction: string;
+}
+
+export interface WorkflowTemplate {
+  id: string;
+  protocol: typeof PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL;
+  version: 1;
+  workflowId: string;
+  name: string;
+  description: string;
+  inputPlaceholder: string;
+  agentSlots: WorkflowTemplateAgentSlot[];
+  acceptanceCriteria: string[];
+  defaultSettings?: {
+    maxParallelAgents?: number;
+    maxRepairCycles?: number;
+  };
+}
+
+type AgentRecord = Record<string, unknown>;
+
+export const builtInWorkflowTemplates: WorkflowTemplate[] = [
+  {
+    id: 'bug_fix_team',
+    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
+    version: 1,
+    workflowId: 'agent_team',
+    name: 'Bug fix team',
+    description: 'Reproduce, fix, and verify a focused bug with an implementation and review loop.',
+    inputPlaceholder: 'Describe the bug, expected behavior, current behavior, and any reproduction steps.',
+    agentSlots: [
+      {
+        id: 'reproducer',
+        role: 'implementation',
+        label: 'Reproducer',
+        instruction: 'Reproduce the bug from the user report, identify the failing path, and keep the scope narrow.',
+      },
+      {
+        id: 'fixer',
+        role: 'implementation',
+        label: 'Fixer',
+        instruction: 'Implement the smallest durable fix and report changed files and verification commands.',
+      },
+      {
+        id: 'reviewer',
+        role: 'review',
+        label: 'Reviewer',
+        instruction: 'Review the fix for regressions, missing validation, and whether the original bug is covered.',
+      },
+    ],
+    acceptanceCriteria: [
+      'The bug has a concrete reproduction or a clearly documented blocker.',
+      'The fix is limited to the failing behavior.',
+      'Verification commands and remaining risks are reported.',
+    ],
+    defaultSettings: {
+      maxParallelAgents: 2,
+      maxRepairCycles: 1,
+    },
+  },
+  {
+    id: 'pr_review_team',
+    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
+    version: 1,
+    workflowId: 'multi_model_review',
+    name: 'PR review team',
+    description: 'Run independent reviewers and aggregate actionable findings for a pull request or diff.',
+    inputPlaceholder: 'Paste the PR link, branch, or diff summary to review.',
+    agentSlots: [
+      {
+        id: 'correctness',
+        role: 'review',
+        label: 'Correctness reviewer',
+        instruction: 'Prioritize correctness bugs, regressions, missing tests, and broken edge cases.',
+      },
+      {
+        id: 'security',
+        role: 'review',
+        label: 'Security reviewer',
+        instruction: 'Prioritize security, secret handling, permission, and data exposure risks.',
+      },
+      {
+        id: 'reporter',
+        role: 'report',
+        label: 'Report aggregator',
+        instruction: 'Aggregate findings by severity with file references and concrete fix suggestions.',
+      },
+    ],
+    acceptanceCriteria: [
+      'Findings are ordered by severity and avoid speculative noise.',
+      'Each issue includes the concrete impact and suggested fix.',
+      'The final report clearly says when no blocking issue is found.',
+    ],
+    defaultSettings: {
+      maxParallelAgents: 3,
+    },
+  },
+  {
+    id: 'frontend_polish',
+    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
+    version: 1,
+    workflowId: 'agent_team',
+    name: 'Frontend polish',
+    description: 'Improve a UI workflow with implementation, responsive checks, and UX review.',
+    inputPlaceholder: 'Describe the screen or workflow that needs frontend polish.',
+    agentSlots: [
+      {
+        id: 'ui_implementation',
+        role: 'frontend',
+        label: 'UI implementer',
+        instruction: 'Implement the requested UI change using existing design conventions and responsive constraints.',
+      },
+      {
+        id: 'ux_review',
+        role: 'review',
+        label: 'UX reviewer',
+        instruction: 'Review layout, text overflow, responsive behavior, accessibility, and interaction states.',
+      },
+      {
+        id: 'verification',
+        role: 'review',
+        label: 'Verification reviewer',
+        instruction: 'Verify the change with available typecheck, lint, build, smoke, or manual UI evidence.',
+      },
+    ],
+    acceptanceCriteria: [
+      'The UI follows existing components and visual density.',
+      'Text and controls do not overlap on mobile or desktop.',
+      'Verification evidence is included in the final summary.',
+    ],
+    defaultSettings: {
+      maxParallelAgents: 2,
+      maxRepairCycles: 1,
+    },
+  },
+  {
+    id: 'release_manager',
+    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
+    version: 1,
+    workflowId: 'sequential_handoff',
+    name: 'Release manager',
+    description: 'Prepare a release checklist, run verification, and summarize publish readiness.',
+    inputPlaceholder: 'Describe the version, release scope, and target channels.',
+    agentSlots: [
+      {
+        id: 'release_plan',
+        role: 'implementation',
+        label: 'Release planner',
+        instruction: 'Create a release checklist, verify version surfaces, and identify publish blockers.',
+      },
+      {
+        id: 'release_verification',
+        role: 'review',
+        label: 'Release verifier',
+        instruction: 'Run or inspect release verification commands and report the exact publish state.',
+      },
+      {
+        id: 'release_report',
+        role: 'report',
+        label: 'Release reporter',
+        instruction: 'Summarize what was released, what was verified, and any blocked channels.',
+      },
+    ],
+    acceptanceCriteria: [
+      'Version surfaces are aligned.',
+      'Build/package verification is reported.',
+      'Remote artifact and publish state are explicitly stated.',
+    ],
+    defaultSettings: {
+      maxParallelAgents: 1,
+    },
+  },
+  {
+    id: 'dependency_audit',
+    protocol: PIXCODE_WORKFLOW_TEMPLATE_PROTOCOL,
+    version: 1,
+    workflowId: 'agent_team',
+    name: 'Dependency audit',
+    description: 'Inspect dependency, runtime, and package risks without making broad upgrades by default.',
+    inputPlaceholder: 'Describe the dependency or runtime area to audit.',
+    agentSlots: [
+      {
+        id: 'audit',
+        role: 'implementation',
+        label: 'Audit analyst',
+        instruction: 'Inspect dependency and runtime risk using existing lockfiles and configured package managers.',
+      },
+      {
+        id: 'risk_review',
+        role: 'review',
+        label: 'Risk reviewer',
+        instruction: 'Validate the findings, call out false positives, and avoid broad upgrade churn unless requested.',
+      },
+      {
+        id: 'report',
+        role: 'report',
+        label: 'Audit reporter',
+        instruction: 'Produce a prioritized remediation plan with commands, risks, and no-secret output.',
+      },
+    ],
+    acceptanceCriteria: [
+      'Findings are tied to actual dependency evidence.',
+      'No unrelated dependency upgrades are proposed as automatic work.',
+      'Security and runtime risks are separated from maintenance suggestions.',
+    ],
+    defaultSettings: {
+      maxParallelAgents: 2,
+    },
+  },
+];
+
+export function getWorkflowTemplate(templateId: string): WorkflowTemplate | undefined {
+  return builtInWorkflowTemplates.find((template) => template.id === templateId);
+}
+
+function applyTemplateSlots(agents: AgentRecord[] | undefined, template: WorkflowTemplate): AgentRecord[] | undefined {
+  if (!Array.isArray(agents) || agents.length === 0) return agents;
+
+  let enabledSlotIndex = 0;
+  return agents.map((agent) => {
+    if (agent.enabled === false) return agent;
+    const slot = template.agentSlots[enabledSlotIndex];
+    enabledSlotIndex += 1;
+    if (!slot) return agent;
+    return {
+      ...agent,
+      role: slot.role,
+      instruction: typeof agent.instruction === 'string' && agent.instruction.trim()
+        ? agent.instruction
+        : slot.instruction,
+      templateSlotId: slot.id,
+      templateSlotLabel: slot.label,
+    };
+  });
+}
+
+export function applyWorkflowTemplateToMetadata(
+  template: WorkflowTemplate,
+  metadata?: Record<string, unknown>,
+): Record<string, unknown> {
+  const settings = metadata?.settings && typeof metadata.settings === 'object'
+    ? metadata.settings as Record<string, unknown>
+    : {};
+
+  return {
+    ...metadata,
+    workflowTemplate: {
+      protocol: template.protocol,
+      id: template.id,
+      version: template.version,
+      name: template.name,
+      workflowId: template.workflowId,
+      acceptanceCriteria: template.acceptanceCriteria,
+      agentSlots: template.agentSlots.map((slot) => ({
+        id: slot.id,
+        role: slot.role,
+        label: slot.label,
+      })),
+    },
+    agents: applyTemplateSlots(metadata?.agents as AgentRecord[] | undefined, template),
+    settings: {
+      ...settings,
+      ...template.defaultSettings,
+    },
+  };
+}