Skip to content

Agent Zero AI framework

Refactor A1/A2… as Internal Tools Within A0

Wabifocus37af2c695195
diff --git a/webui/js/messages.js b/webui/js/messages.js
index 286e23cd6..3368fd922 100644
--- a/webui/js/messages.js
+++ b/webui/js/messages.js
@@ -17,7 +17,17 @@ const PROCESS_TYPES = ['agent', 'tool', 'code_exe', 'browser', 'info', 'hint', '
// Main types that should always be visible (not collapsed)
const MAIN_TYPES = ['user', 'response', 'warning', 'error', 'rate_limit'];
-// Simplified implementation - no complex interactions needed
+/**
+ * Check if a response is from the main agent (A0)
+ * Subordinate agents (A1, A2, ...) responses should be treated as process steps
+ */
+function isMainAgentResponse(heading) {
+ if (!heading) return true; // Default to main agent
+ // Check for subordinate agent patterns like "A1:", "A2:", etc.
+ const match = heading.match(/\bA(\d+):/);
+ if (!match) return true; // No agent marker = main agent
+ return match[1] === "0"; // Only A0 is the main agent
+}
export function setMessage(id, type, heading, content, temp, kvps = null, timestamp = null) {
// Check if this is a process type message
@@ -53,7 +63,25 @@ export function setMessage(id, type, heading, content, temp, kvps = null, timest
return processStepElement;
}
- // For response type, embed the current process group inside (but keep reference)
+ // For subordinate agent responses (A1, A2, ...), treat as a process step instead of main response
+ if (type === "response" && !isMainAgentResponse(heading)) {
+ if (processStepElement) {
+ updateProcessStep(processStepElement, id, "agent", heading, content, kvps);
+ return processStepElement;
+ }
+
+ // Create or get process group for current interaction
+ if (!currentProcessGroup || !document.getElementById(currentProcessGroup.id)) {
+ currentProcessGroup = createProcessGroup(id);
+ chatHistory.appendChild(currentProcessGroup);
+ }
+
+ // Add subordinate response as a step (type "agent" for appropriate styling)
+ processStepElement = addProcessStep(currentProcessGroup, id, "agent", heading, content, kvps, timestamp);
+ return processStepElement;
+ }
+
+ // For main agent (A0) response, embed the current process group
if (type === "response" && currentProcessGroup) {
const processGroupToEmbed = currentProcessGroup;
// Keep currentProcessGroup reference - subsequent process messages go to same group
@@ -1317,7 +1345,8 @@ function getStepTitle(heading, kvps, type) {
}
/**
- * Clean step title by removing icon:// prefixes and agent markers
+ * Clean step title by removing icon:// prefixes
+ * Preserves agent markers (A0:, A1:, etc.) so users can see which agent is executing
*/
function cleanStepTitle(text, maxLength) {
if (!text) return "";
@@ -1326,8 +1355,9 @@ function cleanStepTitle(text, maxLength) {
// Remove icon:// patterns (e.g., "icon://network_intelligence")
cleaned = cleaned.replace(/icon:\/\/[a-zA-Z0-9_]+\s*/g, "");
- // Remove agent markers like "A0:" or similar
- cleaned = cleaned.replace(/^[A-Z]\d+:\s*/i, "");
+ // Keep agent markers (A0:, A1:, etc.) - users need to see which agent is executing
+ // Only remove "A0:" as it's the main agent (implied)
+ cleaned = cleaned.replace(/^A0:\s*/i, "");
// Trim whitespace
cleaned = cleaned.trim();