@humain/terminal 0.0.2 → 0.0.3-beta.0

package/dist/core/agent-session.js

@@ -15,11 +15,12 @@
 import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs";
 import { basename, dirname } from "node:path";
 import { contentText } from "@earendil-works/pi-ai";
-import { clampThinkingLevel, cleanupSessionResources, getSupportedThinkingLevels, isContextOverflow, isRetryableAssistantError, modelsAreEqual, resetApiProviders, streamSimple, } from "@earendil-works/pi-ai/compat";
+import { clampThinkingLevel, cleanupSessionResources, getSupportedThinkingLevels, isContextOverflow, isRecoverableLength, isRetryableAssistantError, modelsAreEqual, resetApiProviders, streamSimple, } from "@earendil-works/pi-ai/compat";
 import { getThemeByName, theme } from "../modes/interactive/theme/theme.js";
 import { stripFrontmatter } from "../utils/frontmatter.js";
 import { resolvePath } from "../utils/paths.js";
 import { sleep } from "../utils/sleep.js";
+import { normalizeToolResultImages } from "../utils/tool-result-images.js";
 import { formatNoApiKeyFoundMessage, formatNoModelSelectedMessage } from "./auth-guidance.js";
 import { executeBashWithOperations } from "./bash-executor.js";
 import { calculateContextTokens, collectEntriesForBranchSummary, compact, estimateContextTokens, estimateTokens, generateBranchSummary, prepareCompaction, shouldCompact, } from "./compaction/index.js";
@@ -175,7 +176,9 @@ export class AgentSession {
             throw error;
         }
         if (result && (result.auth.apiKey || result.auth.headers)) {
+            const requestModel = result.auth.baseUrl ? { ...model, baseUrl: result.auth.baseUrl } : model;
             return {
+                model: requestModel,
                 apiKey: result.auth.apiKey,
                 headers: withoutDeletedHeaders(result.auth.headers),
                 env: result.env,
@@ -195,12 +198,18 @@ export class AgentSession {
         }
         try {
             const result = await this._modelRuntime.getAuth(model);
-            return result
-                ? { apiKey: result.auth.apiKey, headers: withoutDeletedHeaders(result.auth.headers), env: result.env }
-                : {};
+            if (!result)
+                return { model };
+            const requestModel = result.auth.baseUrl ? { ...model, baseUrl: result.auth.baseUrl } : model;
+            return {
+                model: requestModel,
+                apiKey: result.auth.apiKey,
+                headers: withoutDeletedHeaders(result.auth.headers),
+                env: result.env,
+            };
         }
         catch {
-            return {};
+            return { model };
         }
     }
     /**
@@ -234,27 +243,31 @@ export class AgentSession {
         };
         this.agent.afterToolCall = async ({ toolCall, args, result, isError }) => {
             const runner = this._extensionRunner;
-            if (!runner.hasHandlers("tool_result")) {
-                return undefined;
-            }
-            const hookResult = await runner.emitToolResult({
-                type: "tool_result",
-                toolName: toolCall.name,
-                toolCallId: toolCall.id,
-                input: args,
-                content: result.content,
-                details: result.details,
-                isError,
-                usage: result.usage,
+            const hookResult = runner.hasHandlers("tool_result")
+                ? await runner.emitToolResult({
+                    type: "tool_result",
+                    toolName: toolCall.name,
+                    toolCallId: toolCall.id,
+                    input: args,
+                    content: result.content,
+                    details: result.details,
+                    isError,
+                    usage: result.usage,
+                })
+                : undefined;
+            const content = hookResult?.content ?? result.content ?? [];
+            // Runs after the extension hook so images injected or replaced by extensions are normalized too.
+            const normalizedContent = await normalizeToolResultImages(content, {
+                autoResizeImages: this.settingsManager.getImageAutoResize(),
             });
-            if (!hookResult) {
+            if (!hookResult && normalizedContent === content) {
                 return undefined;
             }
             return {
-                content: hookResult.content,
-                details: hookResult.details,
-                isError: hookResult.isError ?? isError,
-                usage: hookResult.usage,
+                content: normalizedContent,
+                details: hookResult?.details,
+                isError: hookResult?.isError ?? isError,
+                usage: hookResult?.usage,
             };
         };
     }
@@ -369,7 +382,7 @@ export class AgentSession {
             if (event.message.role === "assistant") {
                 this._lastAssistantMessage = event.message;
                 const assistantMsg = event.message;
-                if (assistantMsg.stopReason !== "error") {
+                if (assistantMsg.stopReason !== "error" && assistantMsg.stopReason !== "length") {
                     this._overflowRecoveryAttempted = false;
                 }
                 // Reset retry counter immediately on successful assistant response
@@ -529,26 +542,13 @@ export class AgentSession {
             }
         };
     }
-    /**
-     * Temporarily disconnect from agent events.
-     * User listeners are preserved and will receive events again after resubscribe().
-     * Used internally during operations that need to pause event processing.
-     */
+    /** Disconnect from agent events during disposal. */
     _disconnectFromAgent() {
         if (this._unsubscribeAgent) {
             this._unsubscribeAgent();
             this._unsubscribeAgent = undefined;
         }
     }
-    /**
-     * Reconnect to agent events after _disconnectFromAgent().
-     * Preserves all existing listeners.
-     */
-    _reconnectToAgent() {
-        if (this._unsubscribeAgent)
-            return; // Already connected
-        this._unsubscribeAgent = this.agent.subscribe(this._handleAgentEvent);
-    }
     /**
      * Remove all listeners and disconnect from agent.
      * Call this when completely done with the session.
@@ -804,6 +804,9 @@ export class AgentSession {
                     return;
                 }
             }
+            if (this._compactionAbortController !== undefined) {
+                throw new Error("Cannot submit a prompt while compaction is in progress. Wait for compaction to finish and retry.");
+            }
             // Emit input event for extension interception (before skill/template expansion)
             let currentText = text;
             let currentImages = options?.images;
@@ -1217,11 +1220,8 @@ export class AgentSession {
         return this._cycleAvailableModel(direction);
     }
     async _cycleScopedModel(direction) {
-        const checks = await Promise.all(this._scopedModels.map(async (scoped) => ({
-            scoped,
-            auth: await this._modelRuntime.checkAuth(scoped.model.provider),
-        })));
-        const scopedModels = checks.filter(({ auth }) => auth !== undefined).map(({ scoped }) => scoped);
+        const availableIds = new Set(this._modelRuntime.getAvailableSnapshot().map((model) => `${model.provider}\0${model.id}`));
+        const scopedModels = this._scopedModels.filter((scoped) => availableIds.has(`${scoped.model.provider}\0${scoped.model.id}`));
         if (scopedModels.length <= 1)
             return undefined;
         const currentModel = this.model;
@@ -1245,7 +1245,7 @@ export class AgentSession {
         return { model: next.model, thinkingLevel: this.thinkingLevel, isScoped: true };
     }
     async _cycleAvailableModel(direction) {
-        const availableModels = await this._modelRuntime.getAvailable();
+        const availableModels = this._modelRuntime.getAvailableSnapshot();
         if (availableModels.length <= 1)
             return undefined;
         const currentModel = this.model;
@@ -1365,7 +1365,6 @@ export class AgentSession {
      * @param customInstructions Optional instructions for the compaction summary
      */
     async compact(customInstructions) {
-        this._disconnectFromAgent();
         await this.abort();
         this._compactionAbortController = new AbortController();
         this._emit({ type: "compaction_start", reason: "manual" });
@@ -1373,7 +1372,7 @@ export class AgentSession {
             if (!this.model) {
                 throw new Error(formatNoModelSelectedMessage());
             }
-            const { apiKey, headers, env } = await this._getSummarizationRequestAuth(this.model);
+            const { model: requestModel, apiKey, headers, env } = await this._getSummarizationRequestAuth(this.model);
             const pathEntries = this.sessionManager.getBranch();
             const settings = this.settingsManager.getCompactionSettings();
             const preparation = prepareCompaction(pathEntries, settings);
@@ -1420,7 +1419,7 @@ export class AgentSession {
             }
             else {
                 // Generate compaction result
-                const result = await compact(preparation, this.model, apiKey, headers, customInstructions, this._compactionAbortController.signal, this.thinkingLevel, this.agent.streamFunction, env, this.settingsManager.getRetrySettings(), this._summarizationRetryCallbacks({ source: "compaction", reason: "manual" }));
+                const result = await compact(preparation, requestModel, apiKey, headers, customInstructions, this._compactionAbortController.signal, this.thinkingLevel, this.agent.streamFunction, env, this.settingsManager.getRetrySettings(), this._summarizationRetryCallbacks({ source: "compaction", reason: "manual" }));
                 summary = result.summary;
                 firstKeptEntryId = result.firstKeptEntryId;
                 tokensBefore = result.tokensBefore;
@@ -1454,6 +1453,8 @@ export class AgentSession {
                 usage,
                 details,
             };
+            // compaction_end listeners may submit queued prompts, so expose idle state before notifying them.
+            this._compactionAbortController = undefined;
             this._emit({
                 type: "compaction_end",
                 reason: "manual",
@@ -1466,6 +1467,7 @@ export class AgentSession {
         catch (error) {
             const message = error instanceof Error ? error.message : String(error);
             const aborted = message === "Compaction cancelled" || (error instanceof Error && error.name === "AbortError");
+            this._compactionAbortController = undefined;
             this._emit({
                 type: "compaction_end",
                 reason: "manual",
@@ -1478,7 +1480,6 @@ export class AgentSession {
         }
         finally {
             this._compactionAbortController = undefined;
-            this._reconnectToAgent();
         }
     }
     /**
@@ -1499,7 +1500,8 @@ export class AgentSession {
      * Called after agent_end and before prompt submission.
      *
      * Two cases:
-     * 1. Overflow: LLM returned context overflow error, remove error message from agent state, compact, auto-retry
+     * 1. Recoverable failure: LLM returned context overflow or stopped below its desired output limit;
+     *    remove the assistant message, compact, and auto-retry once
      * 2. Threshold: Context over threshold, compact, NO auto-retry (user continues manually)
      *
      * @param assistantMessage The assistant message to check
@@ -1526,11 +1528,13 @@ export class AgentSession {
         if (assistantIsFromBeforeCompaction) {
             return false;
         }
-        // Case 1: Overflow - LLM returned context overflow error, or reported usage exceeded
-        // the configured window. A successful response over the configured window should compact
-        // but must not retry: the assistant answer already completed and agent.continue() cannot
-        // continue from an assistant message.
-        if (sameModel && isContextOverflow(assistantMessage, contextWindow)) {
+        // Case 1: Recoverable failure. Explicit/silent context overflow still uses context metadata.
+        // A length stop is recoverable when output ended below the model's original desired limit,
+        // independent of the configured context size or any context-clamped provider request limit.
+        // A successful response over the configured window should compact but must not retry: the
+        // assistant answer already completed and agent.continue() cannot continue from an assistant.
+        const recoverableLength = sameModel && isRecoverableLength(assistantMessage, this.model?.maxTokens ?? 0);
+        if (sameModel && (isContextOverflow(assistantMessage, contextWindow) || recoverableLength)) {
             const willRetry = assistantMessage.stopReason !== "stop";
             if (!willRetry) {
                 return await this._runAutoCompaction("overflow", false);
@@ -1547,8 +1551,8 @@ export class AgentSession {
                 return false;
             }
             this._overflowRecoveryAttempted = true;
-            // Remove the error message from agent state (it IS saved to session for history,
-            // but we don't want it in context for the retry)
+            // Remove the failed or truncated message from agent state. It remains in session history,
+            // but must not be included in the compact-and-retry context.
             const messages = this.agent.state.messages;
             if (messages.length > 0 && messages[messages.length - 1].role === "assistant") {
                 this.agent.state.messages = messages.slice(0, -1);
@@ -1595,15 +1599,7 @@ export class AgentSession {
             if (!this.model) {
                 return false;
             }
-            let apiKey;
-            let headers;
-            let env;
-            if (this.agent.streamFunction === streamSimple) {
-                ({ apiKey, headers, env } = await this._getRequiredRequestAuth(this.model));
-            }
-            else {
-                ({ apiKey, headers, env } = await this._getSummarizationRequestAuth(this.model));
-            }
+            const { model: requestModel, apiKey, headers, env } = await this._getSummarizationRequestAuth(this.model);
             const pathEntries = this.sessionManager.getBranch();
             const preparation = prepareCompaction(pathEntries, settings);
             if (!preparation) {
@@ -1654,7 +1650,7 @@ export class AgentSession {
             }
             else {
                 // Generate compaction result
-                const compactResult = await compact(preparation, this.model, apiKey, headers, undefined, this._autoCompactionAbortController.signal, this.thinkingLevel, this.agent.streamFunction, env, this.settingsManager.getRetrySettings(), this._summarizationRetryCallbacks({ source: "compaction", reason }));
+                const compactResult = await compact(preparation, requestModel, apiKey, headers, undefined, this._autoCompactionAbortController.signal, this.thinkingLevel, this.agent.streamFunction, env, this.settingsManager.getRetrySettings(), this._summarizationRetryCallbacks({ source: "compaction", reason }));
                 summary = compactResult.summary;
                 firstKeptEntryId = compactResult.firstKeptEntryId;
                 tokensBefore = compactResult.tokensBefore;
@@ -1699,7 +1695,11 @@ export class AgentSession {
             if (willRetry) {
                 const messages = this.agent.state.messages;
                 const lastMsg = messages[messages.length - 1];
-                if (lastMsg?.role === "assistant" && lastMsg.stopReason === "error") {
+                // The overflow response was persisted on message_end before _checkCompaction() removed it
+                // from agent state. Rebuilding state from the new compaction can restore that kept entry,
+                // leaving an assistant as the final message. agent.continue() rejects that state, so remove
+                // the retriable error or truncated-length response again before continuing the interrupted turn.
+                if (lastMsg?.role === "assistant" && (lastMsg.stopReason === "error" || lastMsg.stopReason === "length")) {
                     this.agent.state.messages = messages.slice(0, -1);
                 }
                 return true;
@@ -2050,8 +2050,10 @@ export class AgentSession {
         });
     }
     async reload(options) {
-        const previousFlagValues = this._extensionRunner.getFlagValues();
-        await emitSessionShutdownEvent(this._extensionRunner, { type: "session_shutdown", reason: "reload" });
+        const oldRunner = this._extensionRunner;
+        const previousFlagValues = oldRunner.getFlagValues();
+        await emitSessionShutdownEvent(oldRunner, { type: "session_shutdown", reason: "reload" });
+        oldRunner.invalidate();
         await this.settingsManager.reload();
         this.syncQueueModesFromSettings();
         resetApiProviders();
@@ -2370,10 +2372,10 @@ export class AgentSession {
             let summaryUsage;
             if (options.summarize && entriesToSummarize.length > 0 && !extensionSummary) {
                 const model = this.model;
-                const { apiKey, headers, env } = await this._getSummarizationRequestAuth(model);
+                const { model: requestModel, apiKey, headers, env } = await this._getSummarizationRequestAuth(model);
                 const branchSummarySettings = this.settingsManager.getBranchSummarySettings();
                 const result = await generateBranchSummary(entriesToSummarize, {
-                    model,
+                    model: requestModel,
                     apiKey,
                     headers,
                     env,