Merge remote-tracking branch 'origin/main' into feat/model-cycling-enhancements

package-lock.json

@@ -3847,11 +3847,11 @@
 		},
 		"packages/agent": {
 			"name": "@mariozechner/pi-agent",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"license": "MIT",
 			"dependencies": {
-				"@mariozechner/pi-ai": "^0.8.3",
+				"@mariozechner/pi-ai": "^0.8.4",
-				"@mariozechner/pi-tui": "^0.8.3"
+				"@mariozechner/pi-tui": "^0.8.4"
 			},
 			"devDependencies": {
 				"@types/node": "^24.3.0",
@@ -3881,7 +3881,7 @@
 		},
 		"packages/ai": {
 			"name": "@mariozechner/pi-ai",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"license": "MIT",
 			"dependencies": {
 				"@anthropic-ai/sdk": "^0.61.0",
@@ -3922,12 +3922,12 @@
 		},
 		"packages/coding-agent": {
 			"name": "@mariozechner/pi-coding-agent",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"license": "MIT",
 			"dependencies": {
-				"@mariozechner/pi-agent": "^0.8.3",
+				"@mariozechner/pi-agent": "^0.8.4",
-				"@mariozechner/pi-ai": "^0.8.3",
+				"@mariozechner/pi-ai": "^0.8.4",
-				"@mariozechner/pi-tui": "^0.8.3",
+				"@mariozechner/pi-tui": "^0.8.4",
 				"chalk": "^5.5.0",
 				"diff": "^8.0.2",
 				"glob": "^11.0.3"
@@ -3964,10 +3964,10 @@
 		},
 		"packages/pods": {
 			"name": "@mariozechner/pi",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"license": "MIT",
 			"dependencies": {
-				"@mariozechner/pi-agent": "^0.8.3",
+				"@mariozechner/pi-agent": "^0.8.4",
 				"chalk": "^5.5.0"
 			},
 			"bin": {
@@ -3980,7 +3980,7 @@
 		},
 		"packages/proxy": {
 			"name": "@mariozechner/pi-proxy",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"dependencies": {
 				"@hono/node-server": "^1.14.0",
 				"hono": "^4.6.16"
@@ -3996,7 +3996,7 @@
 		},
 		"packages/tui": {
 			"name": "@mariozechner/pi-tui",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"license": "MIT",
 			"dependencies": {
 				"@types/mime-types": "^2.1.4",
@@ -4031,12 +4031,12 @@
 		},
 		"packages/web-ui": {
 			"name": "@mariozechner/pi-web-ui",
-			"version": "0.8.4",
+			"version": "0.8.5",
 			"license": "MIT",
 			"dependencies": {
 				"@lmstudio/sdk": "^1.5.0",
-				"@mariozechner/pi-ai": "^0.8.3",
+				"@mariozechner/pi-ai": "^0.8.4",
-				"@mariozechner/pi-tui": "^0.8.3",
+				"@mariozechner/pi-tui": "^0.8.4",
 				"docx-preview": "^0.3.7",
 				"jszip": "^3.10.1",
 				"lucide": "^0.544.0",

packages/agent/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-agent",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"description": "General-purpose agent with transport abstraction, state management, and attachment support",
 	"type": "module",
 	"main": "./dist/index.js",
@@ -18,8 +18,8 @@
 		"prepublishOnly": "npm run clean && npm run build"
 	},
 	"dependencies": {
-		"@mariozechner/pi-ai": "^0.8.4",
+		"@mariozechner/pi-ai": "^0.8.5",
-		"@mariozechner/pi-tui": "^0.8.4"
+		"@mariozechner/pi-tui": "^0.8.5"
 	},
 	"keywords": [
 		"ai",

packages/agent/src/agent.ts

@@ -77,6 +77,8 @@ export class Agent {
 	private messageTransformer: (messages: AppMessage[]) => Message[] | Promise<Message[]>;
 	private messageQueue: Array<QueuedMessage<AppMessage>> = [];
 	private queueMode: "all" | "one-at-a-time";
+	private runningPrompt?: Promise<void>;
+	private resolveRunningPrompt?: () => void;
 
 	constructor(opts: AgentOptions) {
 		this._state = { ...this._state, ...opts.initialState };
@@ -148,12 +150,37 @@ export class Agent {
 		this.abortController?.abort();
 	}
 
+	/**
+	 * Returns a promise that resolves when the current prompt completes.
+	 * Returns immediately resolved promise if no prompt is running.
+	 */
+	waitForIdle(): Promise<void> {
+		return this.runningPrompt ?? Promise.resolve();
+	}
+
+	/**
+	 * Clear all messages and state. Call abort() first if a prompt is in flight.
+	 */
+	reset() {
+		this._state.messages = [];
+		this._state.isStreaming = false;
+		this._state.streamMessage = null;
+		this._state.pendingToolCalls = new Set<string>();
+		this._state.error = undefined;
+		this.messageQueue = [];
+	}
+
 	async prompt(input: string, attachments?: Attachment[]) {
 		const model = this._state.model;
 		if (!model) {
 			throw new Error("No model configured");
 		}
 
+		// Set up running prompt tracking
+		this.runningPrompt = new Promise<void>((resolve) => {
+			this.resolveRunningPrompt = resolve;
+		});
+
 		// Build user message with attachments
 		const content: Array<TextContent | ImageContent> = [{ type: "text", text: input }];
 		if (attachments?.length) {
@@ -322,6 +349,9 @@ export class Agent {
 			this._state.streamMessage = null;
 			this._state.pendingToolCalls = new Set<string>();
 			this.abortController = undefined;
+			this.resolveRunningPrompt?.();
+			this.runningPrompt = undefined;
+			this.resolveRunningPrompt = undefined;
 		}
 	}
 

packages/ai/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-ai",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"type": "module",
 	"main": "./dist/index.js",

packages/ai/src/models.generated.ts

@@ -5102,23 +5102,6 @@ export const MODELS = {
 			contextWindow: 131072,
 			maxTokens: 16384,
 		} satisfies Model<"openai-completions">,
-		"meta-llama/llama-3.1-405b-instruct": {
-			id: "meta-llama/llama-3.1-405b-instruct",
-			name: "Meta: Llama 3.1 405B Instruct",
-			api: "openai-completions",
-			provider: "openrouter",
-			baseUrl: "https://openrouter.ai/api/v1",
-			reasoning: false,
-			input: ["text"],
-			cost: {
-				input: 3.5,
-				output: 3.5,
-				cacheRead: 0,
-				cacheWrite: 0,
-			},
-			contextWindow: 130815,
-			maxTokens: 4096,
-		} satisfies Model<"openai-completions">,
 		"meta-llama/llama-3.1-70b-instruct": {
 			id: "meta-llama/llama-3.1-70b-instruct",
 			name: "Meta: Llama 3.1 70B Instruct",
@@ -5136,6 +5119,23 @@ export const MODELS = {
 			contextWindow: 131072,
 			maxTokens: 4096,
 		} satisfies Model<"openai-completions">,
+		"meta-llama/llama-3.1-405b-instruct": {
+			id: "meta-llama/llama-3.1-405b-instruct",
+			name: "Meta: Llama 3.1 405B Instruct",
+			api: "openai-completions",
+			provider: "openrouter",
+			baseUrl: "https://openrouter.ai/api/v1",
+			reasoning: false,
+			input: ["text"],
+			cost: {
+				input: 3.5,
+				output: 3.5,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 130815,
+			maxTokens: 4096,
+		} satisfies Model<"openai-completions">,
 		"mistralai/mistral-nemo": {
 			id: "mistralai/mistral-nemo",
 			name: "Mistral: Mistral Nemo",
@@ -5153,9 +5153,9 @@ export const MODELS = {
 			contextWindow: 131072,
 			maxTokens: 16384,
 		} satisfies Model<"openai-completions">,
-		"openai/gpt-4o-mini-2024-07-18": {
+		"openai/gpt-4o-mini": {
-			id: "openai/gpt-4o-mini-2024-07-18",
+			id: "openai/gpt-4o-mini",
-			name: "OpenAI: GPT-4o-mini (2024-07-18)",
+			name: "OpenAI: GPT-4o-mini",
 			api: "openai-completions",
 			provider: "openrouter",
 			baseUrl: "https://openrouter.ai/api/v1",
@@ -5170,9 +5170,9 @@ export const MODELS = {
 			contextWindow: 128000,
 			maxTokens: 16384,
 		} satisfies Model<"openai-completions">,
-		"openai/gpt-4o-mini": {
+		"openai/gpt-4o-mini-2024-07-18": {
-			id: "openai/gpt-4o-mini",
+			id: "openai/gpt-4o-mini-2024-07-18",
-			name: "OpenAI: GPT-4o-mini",
+			name: "OpenAI: GPT-4o-mini (2024-07-18)",
 			api: "openai-completions",
 			provider: "openrouter",
 			baseUrl: "https://openrouter.ai/api/v1",
@@ -5272,23 +5272,6 @@ export const MODELS = {
 			contextWindow: 128000,
 			maxTokens: 4096,
 		} satisfies Model<"openai-completions">,
-		"openai/gpt-4o-2024-05-13": {
-			id: "openai/gpt-4o-2024-05-13",
-			name: "OpenAI: GPT-4o (2024-05-13)",
-			api: "openai-completions",
-			provider: "openrouter",
-			baseUrl: "https://openrouter.ai/api/v1",
-			reasoning: false,
-			input: ["text", "image"],
-			cost: {
-				input: 5,
-				output: 15,
-				cacheRead: 0,
-				cacheWrite: 0,
-			},
-			contextWindow: 128000,
-			maxTokens: 4096,
-		} satisfies Model<"openai-completions">,
 		"openai/gpt-4o": {
 			id: "openai/gpt-4o",
 			name: "OpenAI: GPT-4o",
@@ -5323,6 +5306,23 @@ export const MODELS = {
 			contextWindow: 128000,
 			maxTokens: 64000,
 		} satisfies Model<"openai-completions">,
+		"openai/gpt-4o-2024-05-13": {
+			id: "openai/gpt-4o-2024-05-13",
+			name: "OpenAI: GPT-4o (2024-05-13)",
+			api: "openai-completions",
+			provider: "openrouter",
+			baseUrl: "https://openrouter.ai/api/v1",
+			reasoning: false,
+			input: ["text", "image"],
+			cost: {
+				input: 5,
+				output: 15,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 128000,
+			maxTokens: 4096,
+		} satisfies Model<"openai-completions">,
 		"meta-llama/llama-3-70b-instruct": {
 			id: "meta-llama/llama-3-70b-instruct",
 			name: "Meta: Llama 3 70B Instruct",
@@ -5442,23 +5442,6 @@ export const MODELS = {
 			contextWindow: 128000,
 			maxTokens: 4096,
 		} satisfies Model<"openai-completions">,
-		"openai/gpt-3.5-turbo-0613": {
-			id: "openai/gpt-3.5-turbo-0613",
-			name: "OpenAI: GPT-3.5 Turbo (older v0613)",
-			api: "openai-completions",
-			provider: "openrouter",
-			baseUrl: "https://openrouter.ai/api/v1",
-			reasoning: false,
-			input: ["text"],
-			cost: {
-				input: 1,
-				output: 2,
-				cacheRead: 0,
-				cacheWrite: 0,
-			},
-			contextWindow: 4095,
-			maxTokens: 4096,
-		} satisfies Model<"openai-completions">,
 		"openai/gpt-4-turbo-preview": {
 			id: "openai/gpt-4-turbo-preview",
 			name: "OpenAI: GPT-4 Turbo Preview",
@@ -5476,6 +5459,23 @@ export const MODELS = {
 			contextWindow: 128000,
 			maxTokens: 4096,
 		} satisfies Model<"openai-completions">,
+		"openai/gpt-3.5-turbo-0613": {
+			id: "openai/gpt-3.5-turbo-0613",
+			name: "OpenAI: GPT-3.5 Turbo (older v0613)",
+			api: "openai-completions",
+			provider: "openrouter",
+			baseUrl: "https://openrouter.ai/api/v1",
+			reasoning: false,
+			input: ["text"],
+			cost: {
+				input: 1,
+				output: 2,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 4095,
+			maxTokens: 4096,
+		} satisfies Model<"openai-completions">,
 		"mistralai/mistral-small": {
 			id: "mistralai/mistral-small",
 			name: "Mistral Small",
@@ -5578,23 +5578,6 @@ export const MODELS = {
 			contextWindow: 8191,
 			maxTokens: 4096,
 		} satisfies Model<"openai-completions">,
-		"openai/gpt-4": {
-			id: "openai/gpt-4",
-			name: "OpenAI: GPT-4",
-			api: "openai-completions",
-			provider: "openrouter",
-			baseUrl: "https://openrouter.ai/api/v1",
-			reasoning: false,
-			input: ["text"],
-			cost: {
-				input: 30,
-				output: 60,
-				cacheRead: 0,
-				cacheWrite: 0,
-			},
-			contextWindow: 8191,
-			maxTokens: 4096,
-		} satisfies Model<"openai-completions">,
 		"openai/gpt-3.5-turbo": {
 			id: "openai/gpt-3.5-turbo",
 			name: "OpenAI: GPT-3.5 Turbo",
@@ -5612,6 +5595,23 @@ export const MODELS = {
 			contextWindow: 16385,
 			maxTokens: 4096,
 		} satisfies Model<"openai-completions">,
+		"openai/gpt-4": {
+			id: "openai/gpt-4",
+			name: "OpenAI: GPT-4",
+			api: "openai-completions",
+			provider: "openrouter",
+			baseUrl: "https://openrouter.ai/api/v1",
+			reasoning: false,
+			input: ["text"],
+			cost: {
+				input: 30,
+				output: 60,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 8191,
+			maxTokens: 4096,
+		} satisfies Model<"openai-completions">,
 		"openrouter/auto": {
 			id: "openrouter/auto",
 			name: "OpenRouter: Auto Router",

packages/coding-agent/CHANGELOG.md

@@ -2,6 +2,21 @@
 
 ## [Unreleased]
 
+### Added
+
+- **`/clear` Command**: New slash command to reset the conversation context and start a fresh session. Aborts any in-flight agent work, clears all messages, and creates a new session file. ([#48](https://github.com/badlogic/pi-mono/pull/48))
+
+### Fixed
+
+- **Markdown Link Rendering**: Fixed links with identical text and href (e.g., `https://github.com/badlogic/pi-mono/pull/48/files`) being rendered twice. Now correctly compares raw text instead of styled text (which contains ANSI codes) when determining if link text matches href.
+
+## [0.8.5] - 2025-11-21
+
+### Fixed
+
+- **Path Completion Hanging**: Fixed catastrophic regex backtracking in path completion that caused the terminal to hang when text contained many `/` characters (e.g., URLs). Replaced complex regex with simple string operations. ([#18](https://github.com/badlogic/pi-mono/issues/18))
+- **Autocomplete Arrow Keys**: Fixed issue where arrow keys would move both the autocomplete selection and the editor cursor simultaneously when the file selector list was shown.
+
 ## [0.8.4] - 2025-11-21
 
 ### Fixed

packages/coding-agent/README.md

@@ -445,6 +445,16 @@ Logout from OAuth providers:
 
 Shows a list of logged-in providers to logout from.
 
+### /clear
+
+Clear the conversation context and start a fresh session:
+
+```
+/clear
+```
+
+Aborts any in-flight agent work, clears all messages, and creates a new session file.
+
 ## Editor Features
 
 The interactive input editor includes several productivity features:

packages/coding-agent/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-coding-agent",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"description": "Coding agent CLI with read, bash, edit, write tools and session management",
 	"type": "module",
 	"bin": {
@@ -22,9 +22,9 @@
 		"prepublishOnly": "npm run clean && npm run build"
 	},
 	"dependencies": {
-		"@mariozechner/pi-agent": "^0.8.4",
+		"@mariozechner/pi-agent": "^0.8.5",
-		"@mariozechner/pi-ai": "^0.8.4",
+		"@mariozechner/pi-ai": "^0.8.5",
-		"@mariozechner/pi-tui": "^0.8.4",
+		"@mariozechner/pi-tui": "^0.8.5",
 		"chalk": "^5.5.0",
 		"diff": "^8.0.2",
 		"glob": "^11.0.3"

packages/coding-agent/src/main.ts

@@ -21,17 +21,6 @@ const __dirname = dirname(__filename);
 const packageJson = JSON.parse(readFileSync(join(__dirname, "../package.json"), "utf-8"));
 const VERSION = packageJson.version;
 
-const envApiKeyMap: Record<KnownProvider, string[]> = {
-	google: ["GEMINI_API_KEY"],
-	openai: ["OPENAI_API_KEY"],
-	anthropic: ["ANTHROPIC_OAUTH_TOKEN", "ANTHROPIC_API_KEY"],
-	xai: ["XAI_API_KEY"],
-	groq: ["GROQ_API_KEY"],
-	cerebras: ["CEREBRAS_API_KEY"],
-	openrouter: ["OPENROUTER_API_KEY"],
-	zai: ["ZAI_API_KEY"],
-};
-
 const defaultModelPerProvider: Record<KnownProvider, string> = {
 	anthropic: "claude-sonnet-4-5",
 	openai: "gpt-5.1-codex",
@@ -478,14 +467,9 @@ async function runInteractiveMode(
 		scopedModels,
 	);
 
-	// Initialize TUI
+	// Initialize TUI (subscribes to agent events internally)
 	await renderer.init();
 
-	// Set interrupt callback
-	renderer.setInterruptCallback(() => {
-		agent.abort();
-	});
-
 	// Render any existing messages (from --continue mode)
 	renderer.renderInitialMessages(agent.state);
 
@@ -494,12 +478,6 @@ async function runInteractiveMode(
 		renderer.showWarning(modelFallbackMessage);
 	}
 
-	// Subscribe to agent events
-	agent.subscribe(async (event) => {
-		// Pass all events to the renderer
-		await renderer.handleEvent(event, agent.state);
-	});
-
 	// Interactive loop
 	while (true) {
 		const userInput = await renderer.getUserInput();
@@ -718,11 +696,6 @@ export async function main(args: string[]) {
 	// Load previous messages if continuing or resuming
 	// This may update initialModel if restoring from session
 	if (parsed.continue || parsed.resume) {
-		const messages = sessionManager.loadMessages();
-		if (messages.length > 0 && shouldPrintMessages) {
-			console.log(chalk.dim(`Loaded ${messages.length} messages from previous session`));
-		}
-
 		// Load and restore model (overrides initialModel if found and has API key)
 		const savedModel = sessionManager.loadModel();
 		if (savedModel) {
@@ -871,9 +844,6 @@ export async function main(args: string[]) {
 		}
 	}
 
-	// Note: Session will be started lazily after first user+assistant message exchange
-	// (unless continuing/resuming, in which case it's already initialized)
-
 	// Log loaded context files (they're already in the system prompt)
 	if (shouldPrintMessages && !parsed.continue && !parsed.resume) {
 		const contextFiles = loadProjectContextFiles();
@@ -885,19 +855,6 @@ export async function main(args: string[]) {
 		}
 	}
 
-	// Subscribe to agent events to save messages
-	agent.subscribe((event) => {
-		// Save messages on completion
-		if (event.type === "message_end") {
-			sessionManager.saveMessage(event.message);
-
-			// Check if we should initialize session now (after first user+assistant exchange)
-			if (sessionManager.shouldInitializeSession(agent.state.messages)) {
-				sessionManager.startSession(agent.state);
-			}
-		}
-	});
-
 	// Route to appropriate mode
 	if (mode === "rpc") {
 		// RPC mode - headless operation
@@ -930,8 +887,6 @@ export async function main(args: string[]) {
 				}
 			} else {
 				// Parse current and last versions
-				const currentParts = VERSION.split(".").map(Number);
-				const current = { major: currentParts[0] || 0, minor: currentParts[1] || 0, patch: currentParts[2] || 0 };
 				const changelogPath = getChangelogPath();
 				const entries = parseChangelog(changelogPath);
 				const newEntries = getNewEntries(entries, lastVersion);

packages/coding-agent/src/session-manager.ts

@@ -97,6 +97,13 @@ export class SessionManager {
 		this.sessionFile = join(this.sessionDir, `${timestamp}_${this.sessionId}.jsonl`);
 	}
 
+	/** Reset to a fresh session. Clears pending messages and starts a new session file. */
+	reset(): void {
+		this.pendingMessages = [];
+		this.sessionInitialized = false;
+		this.initNewSession();
+	}
+
 	private findMostRecentlyModifiedSession(): string | null {
 		try {
 			const files = readdirSync(this.sessionDir)

packages/coding-agent/src/tui/tui-renderer.ts

@@ -53,7 +53,7 @@ export class TuiRenderer {
 	private isInitialized = false;
 	private onInputCallback?: (text: string) => void;
 	private loadingAnimation: Loader | null = null;
-	private onInterruptCallback?: () => void;
+
 	private lastSigintTime = 0;
 	private changelogMarkdown: string | null = null;
 	private newVersion: string | null = null;
@@ -97,6 +97,9 @@ export class TuiRenderer {
 	// Tool output expansion state
 	private toolOutputExpanded = false;
 
+	// Agent subscription unsubscribe function
+	private unsubscribe?: () => void;
+
 	constructor(
 		agent: Agent,
 		sessionManager: SessionManager,
@@ -173,6 +176,11 @@ export class TuiRenderer {
 			description: "Select color theme (opens selector UI)",
 		};
 
+		const clearCommand: SlashCommand = {
+			name: "clear",
+			description: "Clear context and start a fresh session",
+		};
+
 		// Setup autocomplete for file paths and slash commands
 		const autocompleteProvider = new CombinedAutocompleteProvider(
 			[
@@ -186,6 +194,7 @@ export class TuiRenderer {
 				loginCommand,
 				logoutCommand,
 				queueCommand,
+				clearCommand,
 			],
 			process.cwd(),
 		);
@@ -268,7 +277,7 @@ export class TuiRenderer {
 		// Set up custom key handlers on the editor
 		this.editor.onEscape = () => {
 			// Intercept Escape key when processing
-			if (this.loadingAnimation && this.onInterruptCallback) {
+			if (this.loadingAnimation) {
 				// Get all queued messages
 				const queuedText = this.queuedMessages.join("\n\n");
 
@@ -289,7 +298,7 @@ export class TuiRenderer {
 				this.agent.clearMessageQueue();
 
 				// Abort
-				this.onInterruptCallback();
+				this.agent.abort();
 			}
 		};
 
@@ -386,6 +395,13 @@ export class TuiRenderer {
 				return;
 			}
 
+			// Check for /clear command
+			if (text === "/clear") {
+				this.handleClearCommand();
+				this.editor.setText("");
+				return;
+			}
+
 			// Normal message submission - validate model and API key first
 			const currentModel = this.agent.state.model;
 			if (!currentModel) {
@@ -439,6 +455,9 @@ export class TuiRenderer {
 		this.ui.start();
 		this.isInitialized = true;
 
+		// Subscribe to agent events for UI updates and session saving
+		this.subscribeToAgent();
+
 		// Set up theme file watcher for live reload
 		onThemeChange(() => {
 			this.ui.invalidate();
@@ -447,7 +466,24 @@ export class TuiRenderer {
 		});
 	}
 
-	async handleEvent(event: AgentEvent, state: AgentState): Promise<void> {
+	private subscribeToAgent(): void {
+		this.unsubscribe = this.agent.subscribe(async (event) => {
+			// Handle UI updates
+			await this.handleEvent(event, this.agent.state);
+
+			// Save messages to session
+			if (event.type === "message_end") {
+				this.sessionManager.saveMessage(event.message);
+
+				// Check if we should initialize session now (after first user+assistant exchange)
+				if (this.sessionManager.shouldInitializeSession(this.agent.state.messages)) {
+					this.sessionManager.startSession(this.agent.state);
+				}
+			}
+		});
+	}
+
+	private async handleEvent(event: AgentEvent, state: AgentState): Promise<void> {
 		if (!this.isInitialized) {
 			await this.init();
 		}
@@ -713,10 +749,6 @@ export class TuiRenderer {
 		});
 	}
 
-	setInterruptCallback(callback: () => void): void {
-		this.onInterruptCallback = callback;
-	}
-
 	private handleCtrlC(): void {
 		// Handle Ctrl+C double-press logic
 		const now = Date.now();
@@ -1435,6 +1467,45 @@ export class TuiRenderer {
 		this.ui.requestRender();
 	}
 
+	private async handleClearCommand(): Promise<void> {
+		// Unsubscribe first to prevent processing abort events
+		this.unsubscribe?.();
+
+		// Abort and wait for completion
+		this.agent.abort();
+		await this.agent.waitForIdle();
+
+		// Stop loading animation
+		if (this.loadingAnimation) {
+			this.loadingAnimation.stop();
+			this.loadingAnimation = null;
+		}
+		this.statusContainer.clear();
+
+		// Reset agent and session
+		this.agent.reset();
+		this.sessionManager.reset();
+
+		// Resubscribe to agent
+		this.subscribeToAgent();
+
+		// Clear UI state
+		this.chatContainer.clear();
+		this.pendingMessagesContainer.clear();
+		this.queuedMessages = [];
+		this.streamingComponent = null;
+		this.pendingTools.clear();
+		this.isFirstUserMessage = true;
+
+		// Show confirmation
+		this.chatContainer.addChild(new Spacer(1));
+		this.chatContainer.addChild(
+			new Text(theme.fg("accent", "✓ Context cleared") + "\n" + theme.fg("muted", "Started fresh session"), 1, 1),
+		);
+
+		this.ui.requestRender();
+	}
+
 	private updatePendingMessagesDisplay(): void {
 		this.pendingMessagesContainer.clear();
 

packages/pods/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"description": "CLI tool for managing vLLM deployments on GPU pods",
 	"type": "module",
 	"bin": {
@@ -34,7 +34,7 @@
 		"node": ">=20.0.0"
 	},
 	"dependencies": {
-		"@mariozechner/pi-agent": "^0.8.4",
+		"@mariozechner/pi-agent": "^0.8.5",
 		"chalk": "^5.5.0"
 	},
 	"devDependencies": {}

packages/proxy/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-proxy",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"type": "module",
 	"description": "CORS and authentication proxy for pi-ai",
 	"main": "dist/index.js",

packages/tui/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-tui",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"description": "Terminal User Interface library with differential rendering for efficient text-based applications",
 	"type": "module",
 	"main": "dist/index.js",

packages/tui/src/autocomplete.ts

@@ -283,21 +283,17 @@ export class CombinedAutocompleteProvider implements AutocompleteProvider {
 			return atMatch[0]; // Return the full @path pattern
 		}
 
-		// Match paths - including those ending with /, ~/, or any word at end for forced extraction
+		// Simple approach: find the last whitespace/delimiter and extract the word after it
-		// This regex captures:
+		// This avoids catastrophic backtracking from nested quantifiers
-		// - Paths starting from beginning of line or after space/quote/equals
+		const lastDelimiterIndex = Math.max(
-		// - Absolute paths starting with /
+			text.lastIndexOf(" "),
-		// - Relative paths with ./ or ../
+			text.lastIndexOf("\t"),
-		// - Home directory paths with ~/
+			text.lastIndexOf('"'),
-		// - The path itself (can include / in the middle)
+			text.lastIndexOf("'"),
-		// - For forced extraction, capture any word at the end
+			text.lastIndexOf("="),
-		const matches = text.match(/(?:^|[\s"'=])((?:\/|~\/|\.{1,2}\/)?(?:[^\s"'=]*\/?)*[^\s"'=]*)$/);
+		);
-		if (!matches) {
-			// If forced extraction and no matches, return empty string to trigger from current dir
-			return forceExtract ? "" : null;
-		}
 
-		const pathPrefix = matches[1] || "";
+		const pathPrefix = lastDelimiterIndex === -1 ? text : text.slice(lastDelimiterIndex + 1);
 
 		// For forced extraction (Tab key), always return something
 		if (forceExtract) {

packages/tui/src/components/editor.ts

@@ -192,6 +192,7 @@ export class Editor implements Component {
 				// Only pass arrow keys to the list, not Enter/Tab (we handle those directly)
 				if (data === "\x1b[A" || data === "\x1b[B") {
 					this.autocompleteList.handleInput(data);
+					return;
 				}
 
 				// If Tab was pressed, always apply the selection
@@ -832,6 +833,11 @@ export class Editor implements Component {
 		this.tryTriggerAutocomplete(true);
 	}
 
+	/*
+https://github.com/EsotericSoftware/spine-runtimes/actions/runs/19536643416/job/559322883
+17 this job fails with https://github.com/EsotericSoftware/spine-runtimes/actions/runs/19
+536643416/job/55932288317 havea  look at .gi
+	 */
 	private forceFileAutocomplete(): void {
 		if (!this.autocompleteProvider) return;
 

packages/tui/src/components/markdown.ts

@@ -321,7 +321,8 @@ export class Markdown implements Component {
 				case "link": {
 					const linkText = this.renderInlineTokens(token.tokens || []);
 					// If link text matches href, only show the link once
-					if (linkText === token.href) {
+					// Compare raw text (token.text) not styled text (linkText) since linkText has ANSI codes
+					if (token.text === token.href) {
 						result += this.theme.link(this.theme.underline(linkText)) + this.applyDefaultStyle("");
 					} else {
 						result +=

packages/web-ui/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-web-ui",
-	"version": "0.8.4",
+	"version": "0.8.5",
 	"description": "Reusable web UI components for AI chat interfaces powered by @mariozechner/pi-ai",
 	"type": "module",
 	"main": "dist/index.js",
@@ -18,8 +18,8 @@
 	},
 	"dependencies": {
 		"@lmstudio/sdk": "^1.5.0",
-		"@mariozechner/pi-ai": "^0.8.4",
+		"@mariozechner/pi-ai": "^0.8.5",
-		"@mariozechner/pi-tui": "^0.8.4",
+		"@mariozechner/pi-tui": "^0.8.5",
 		"docx-preview": "^0.3.7",
 		"jszip": "^3.10.1",
 		"lucide": "^0.544.0",