@eminent337/aery 0.1.115 → 0.1.117

package/dist/core/agent-session.js

@@ -14,7 +14,7 @@
  */
 import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs";
 import { basename, dirname, resolve } from "node:path";
-import { isContextOverflow, modelsAreEqual, resetApiProviders, supportsXhigh } from "@eminent337/aery-ai";
+import { clampThinkingLevel, cleanupSessionResources, getSupportedThinkingLevels, isContextOverflow, modelsAreEqual, resetApiProviders, } from "@eminent337/aery-ai";
 import { theme } from "../modes/interactive/theme/theme.js";
 import { stripFrontmatter } from "../utils/frontmatter.js";
 import { sleep } from "../utils/sleep.js";
@@ -53,8 +53,6 @@ export function parseSkillBlock(text) {
 // ============================================================================
 /** Standard thinking levels */
 const THINKING_LEVELS = ["off", "minimal", "low", "medium", "high"];
-/** Thinking levels including xhigh (for supported models) */
-const THINKING_LEVELS_WITH_XHIGH = ["off", "minimal", "low", "medium", "high", "xhigh"];
 // ============================================================================
 // AgentSession Class
 // ============================================================================
@@ -113,7 +111,7 @@ export class AgentSession {
     _toolPromptGuidelines = new Map();
     // Base system prompt (without extension appends) - used to apply fresh appends each turn
     _baseSystemPrompt = "";
-    _baseSystemPromptOptions = { cwd: process.cwd() };
+    _baseSystemPromptOptions;
     constructor(config) {
         this.agent = config.agent;
         this.sessionManager = config.sessionManager;
@@ -372,6 +370,20 @@ export class AgentSession {
         }
         return undefined;
     }
+    _replaceMessageInPlace(target, replacement) {
+        // Agent-core stores the finalized message object in its state before emitting message_end.
+        // SessionManager persistence happens later in _processAgentEvent() with event.message.
+        // Mutating this object in place keeps agent state, later turn/agent events, listeners,
+        // and the eventual SessionManager.appendMessage(event.message) persistence in sync.
+        if (target === replacement) {
+            return;
+        }
+        const targetRecord = target;
+        for (const key of Object.keys(targetRecord)) {
+            delete targetRecord[key];
+        }
+        Object.assign(targetRecord, replacement);
+    }
     /** Emit extension events based on agent events */
     async _emitExtensionEvent(event) {
         if (event.type === "agent_start") {
@@ -419,7 +431,10 @@ export class AgentSession {
                 type: "message_end",
                 message: event.message,
             };
-            await this._extensionRunner.emit(extensionEvent);
+            const replacement = await this._extensionRunner.emitMessageEnd(extensionEvent);
+            if (replacement) {
+                this._replaceMessageInPlace(event.message, replacement);
+            }
         }
         else if (event.type === "tool_execution_start") {
             const extensionEvent = {
@@ -494,6 +509,7 @@ export class AgentSession {
         this._extensionRunner.invalidate("This extension ctx is stale after session replacement or reload. Do not use a captured pi or command ctx after ctx.newSession(), ctx.fork(), ctx.switchSession(), or ctx.reload(). For newSession, fork, and switchSession, move post-replacement work into withSession and use the ctx passed to withSession. For reload, do not use the old ctx after await ctx.reload().");
         this._disconnectFromAgent();
         this._eventListeners = [];
+        cleanupSessionResources(this.sessionId);
     }
     // =========================================================================
     // Read-only State Access
@@ -664,7 +680,7 @@ export class AgentSession {
     // =========================================================================
     /**
      * Send a prompt to the agent.
-     * - Handles extension commands (registered via aery.registerCommand) immediately, even during streaming
+     * - Handles extension commands (registered via pi.registerCommand) immediately, even during streaming
      * - Expands file-based prompt templates by default
      * - During streaming, queues via steer() or followUp() based on streamingBehavior option
      * - Validates model and API key before sending (when not streaming)
@@ -1142,13 +1158,20 @@ export class AgentSession {
         const availableLevels = this.getAvailableThinkingLevels();
         const effectiveLevel = availableLevels.includes(level) ? level : this._clampThinkingLevel(level, availableLevels);
         // Only persist if actually changing
-        const isChanging = effectiveLevel !== this.agent.state.thinkingLevel;
+        const previousLevel = this.agent.state.thinkingLevel;
+        const isChanging = effectiveLevel !== previousLevel;
         this.agent.state.thinkingLevel = effectiveLevel;
         if (isChanging) {
             this.sessionManager.appendThinkingLevelChange(effectiveLevel);
             if (this.supportsThinking() || effectiveLevel !== "off") {
                 this.settingsManager.setDefaultThinkingLevel(effectiveLevel);
             }
+            this._emit({ type: "thinking_level_changed", level: effectiveLevel });
+            void this._extensionRunner.emit({
+                type: "thinking_level_select",
+                level: effectiveLevel,
+                previousLevel,
+            });
         }
     }
     /**
@@ -1170,15 +1193,9 @@ export class AgentSession {
      * The provider will clamp to what the specific model supports internally.
      */
     getAvailableThinkingLevels() {
-        if (!this.supportsThinking())
-            return ["off"];
-        return this.supportsXhighThinking() ? THINKING_LEVELS_WITH_XHIGH : THINKING_LEVELS;
-    }
-    /**
-     * Check if current model supports xhigh thinking level.
-     */
-    supportsXhighThinking() {
-        return this.model ? supportsXhigh(this.model) : false;
+        if (!this.model)
+            return THINKING_LEVELS;
+        return getSupportedThinkingLevels(this.model);
     }
     /**
      * Check if current model supports thinking/reasoning.
@@ -1195,24 +1212,8 @@ export class AgentSession {
         }
         return this.thinkingLevel;
     }
-    _clampThinkingLevel(level, availableLevels) {
-        const ordered = THINKING_LEVELS_WITH_XHIGH;
-        const available = new Set(availableLevels);
-        const requestedIndex = ordered.indexOf(level);
-        if (requestedIndex === -1) {
-            return availableLevels[0] ?? "off";
-        }
-        for (let i = requestedIndex; i < ordered.length; i++) {
-            const candidate = ordered[i];
-            if (available.has(candidate))
-                return candidate;
-        }
-        for (let i = requestedIndex - 1; i >= 0; i--) {
-            const candidate = ordered[i];
-            if (available.has(candidate))
-                return candidate;
-        }
-        return availableLevels[0] ?? "off";
+    _clampThinkingLevel(level, _availableLevels) {
+        return this.model ? clampThinkingLevel(this.model, level) : "off";
     }
     // =========================================================================
     // Queue Mode Management
@@ -1926,8 +1927,8 @@ export class AgentSession {
         if (isContextOverflow(message, contextWindow))
             return false;
         const err = message.errorMessage;
-        // Match: overloaded_error, provider returned error, rate limit, 429, 500, 502, 503, 504, service unavailable, network/connection errors (including connection lost), WebSocket transport closes/errors, fetch failed, request ended without sending chunks, HTTP/2 closed before response, terminated, retry delay exceeded
-        return /overloaded|provider.?returned.?error|rate.?limit|too many requests|429|500|502|503|504|service.?unavailable|server.?error|internal.?error|network.?error|connection.?error|connection.?refused|connection.?lost|websocket.?closed|websocket.?error|other side closed|fetch failed|upstream.?connect|reset before headers|socket hang up|ended without|http2 request did not get a response|timed? out|timeout|terminated|retry delay/i.test(err);
+        // Match: overloaded_error, provider returned error, rate limit, 429, 500, 502, 503, 504, service unavailable, network/connection errors (including connection lost), WebSocket transport closes/errors, fetch failed, premature stream endings, HTTP/2 closed before response, terminated, retry delay exceeded
+        return /overloaded|provider.?returned.?error|rate.?limit|too many requests|429|500|502|503|504|service.?unavailable|server.?error|internal.?error|network.?error|connection.?error|connection.?refused|connection.?lost|websocket.?closed|websocket.?error|other side closed|fetch failed|upstream.?connect|reset before headers|socket hang up|ended without|stream ended before message_stop|http2 request did not get a response|timed? out|timeout|terminated|retry delay/i.test(err);
     }
     /**
      * Handle retryable errors with exponential backoff.