diff --git a/src/resources/extensions/genai-proxy/extension-manifest.json b/src/resources/extensions/genai-proxy/extension-manifest.json
deleted file mode 100644
index 11794882d..000000000
--- a/src/resources/extensions/genai-proxy/extension-manifest.json
+++ /dev/null
@@ -1,10 +0,0 @@
-{
-	"id": "genai-proxy",
-	"name": "GenAI Proxy",
-	"version": "1.0.0",
-	"description": "OpenAI-compatible proxy for GenAI clients",
-	"tier": "bundled",
-	"requires": {
-		"platform": ">=2.29.0"
-	}
-}
diff --git a/src/resources/extensions/genai-proxy/index.js b/src/resources/extensions/genai-proxy/index.js
deleted file mode 100644
index 4ee9acf40..000000000
--- a/src/resources/extensions/genai-proxy/index.js
+++ /dev/null
@@ -1,10 +0,0 @@
-import { installGenaiProxyExtension } from "./proxy-command.js";
-
-export {
-	installGenaiProxyExtension,
-	resolveProxyPort,
-} from "./proxy-command.js";
-export { createProxyServer, ProxyServer } from "./proxy-server.js";
-export default function genaiProxyExtension(api) {
-	installGenaiProxyExtension(api);
-}
diff --git a/src/resources/extensions/genai-proxy/package.json b/src/resources/extensions/genai-proxy/package.json
deleted file mode 100644
index 6a5cc661a..000000000
--- a/src/resources/extensions/genai-proxy/package.json
+++ /dev/null
@@ -1,14 +0,0 @@
-{
-	"name": "pi-genai-proxy",
-	"private": true,
-	"version": "1.0.0",
-	"type": "module",
-	"engines": {
-		"node": ">=26.1.0"
-	},
-	"pi": {
-		"extensions": [
-			"./index.js"
-		]
-	}
-}
diff --git a/src/resources/extensions/genai-proxy/proxy-command.js b/src/resources/extensions/genai-proxy/proxy-command.js
deleted file mode 100644
index 21698554c..000000000
--- a/src/resources/extensions/genai-proxy/proxy-command.js
+++ /dev/null
@@ -1,122 +0,0 @@
-import { createProxyServer } from "./proxy-server.js";
-
-const PROXY_COMMAND_NAME = "genai-proxy";
-const PROXY_FLAG_NAME = "genai-proxy";
-const DEFAULT_PROXY_PORT = 3000;
-export function installGenaiProxyExtension(api, dependencies) {
-	let proxyServer = null;
-	const buildProxyServer = dependencies?.createProxyServer ?? createProxyServer;
-	const ensureProxyServer = (context, port) => {
-		if (proxyServer && proxyServer.getPort() === port) {
-			return proxyServer;
-		}
-		if (proxyServer) {
-			throw new Error(`Proxy already running on port ${proxyServer.getPort()}`);
-		}
-		proxyServer = buildProxyServer({
-			port,
-			modelRegistry: context.modelRegistry,
-			onLog: (message) => notifyProxyStatus(context, message, "info"),
-		});
-		return proxyServer;
-	};
-	const startProxyFromFlag = async (value, context) => {
-		const server = ensureProxyServer(context, resolveProxyPort(value));
-		await server.start();
-	};
-	api.registerFlag(PROXY_FLAG_NAME, {
-		description: "Start the GenAI proxy server",
-		type: "string",
-		allowNoValue: true,
-		onStartup: startProxyFromFlag,
-	});
-	api.registerCommand(PROXY_COMMAND_NAME, {
-		description: "Manage the GenAI proxy server",
-		handler: async (args, context) => {
-			await handleProxyCommand(
-				args ?? "",
-				context,
-				ensureProxyServer,
-				() => proxyServer,
-				() => {
-					proxyServer = null;
-				},
-			);
-		},
-	});
-}
-export function resolveProxyPort(flagValue) {
-	if (flagValue === true || flagValue === false || flagValue === undefined) {
-		return DEFAULT_PROXY_PORT;
-	}
-	const port = Number.parseInt(flagValue, 10);
-	if (!Number.isFinite(port) || port <= 0 || port > 65535) {
-		throw new Error(`Invalid proxy port: ${flagValue}`);
-	}
-	return port;
-}
-async function handleProxyCommand(
-	rawArgs,
-	context,
-	ensureProxyServer,
-	getProxyServer,
-	clearProxyServer,
-) {
-	const [subcommand = "status", portArg] = rawArgs
-		.trim()
-		.split(/\s+/)
-		.filter((value) => value.length > 0);
-	if (subcommand === "start") {
-		const existingServer = getProxyServer();
-		if (existingServer?.isRunning()) {
-			notifyProxyStatus(
-				context,
-				`Proxy already running on port ${existingServer.getPort()}`,
-				"info",
-			);
-			return;
-		}
-		const server = ensureProxyServer(
-			context,
-			resolveProxyPort(portArg === undefined ? true : portArg),
-		);
-		await server.start();
-		return;
-	}
-	if (subcommand === "stop") {
-		const server = getProxyServer();
-		if (!server?.isRunning()) {
-			notifyProxyStatus(context, "Proxy is not running", "warning");
-			return;
-		}
-		await server.stop();
-		clearProxyServer();
-		notifyProxyStatus(context, "Proxy stopped", "success");
-		return;
-	}
-	if (subcommand === "status") {
-		const server = getProxyServer();
-		if (server?.isRunning()) {
-			notifyProxyStatus(
-				context,
-				`Proxy running on port ${server.getPort()}`,
-				"info",
-			);
-			return;
-		}
-		notifyProxyStatus(context, "Proxy is not running", "info");
-		return;
-	}
-	notifyProxyStatus(
-		context,
-		"Usage: /genai-proxy start [port] | stop | status",
-		"warning",
-	);
-}
-function notifyProxyStatus(context, message, type) {
-	if ("ui" in context) {
-		context.ui.notify(message, type);
-		return;
-	}
-	process.stderr.write(`[genai-proxy] ${message}\n`);
-}
diff --git a/src/resources/extensions/genai-proxy/proxy-server.js b/src/resources/extensions/genai-proxy/proxy-server.js
deleted file mode 100644
index acc347259..000000000
--- a/src/resources/extensions/genai-proxy/proxy-server.js
+++ /dev/null
@@ -1,336 +0,0 @@
-import { stream } from "@singularity-forge/pi-ai";
-import express from "express";
-
-const LISTEN_ADDRESS = "127.0.0.1";
-const OPENAI_CREATED_TIMESTAMP = 1_677_610_602;
-const SSE_CONTENT_TYPE = "text/event-stream";
-const NDJSON_CONTENT_TYPE = "application/x-ndjson";
-export class ProxyServer {
-	server = null;
-	boundPort = null;
-	options;
-	streamModel;
-	constructor(options) {
-		this.options = options;
-		this.streamModel = options.streamModel ?? stream;
-	}
-	isRunning() {
-		return this.server !== null;
-	}
-	getPort() {
-		return this.boundPort;
-	}
-	async start() {
-		if (this.server) {
-			return;
-		}
-		const app = express();
-		app.use(express.json({ limit: "2mb" }));
-		app.get(["/v1/models", "/v1beta/models"], (_req, res) => {
-			const models = this.options.modelRegistry.getAll().map((model) => ({
-				id: model.id,
-				object: "model",
-				created: OPENAI_CREATED_TIMESTAMP,
-				owned_by: model.provider,
-				name: model.name,
-				capabilities: model.capabilities,
-			}));
-			if (_req.path.startsWith("/v1beta")) {
-				res.json({ models });
-				return;
-			}
-			res.json({ object: "list", data: models });
-		});
-		app.post("/v1/chat/completions", async (req, res) => {
-			await this.handleCompletionRequest(req, res, "openai");
-		});
-		app.post(
-			"/v1beta/models/:modelId\\:streamGenerateContent",
-			async (req, res) => {
-				await this.handleCompletionRequest(req, res, "google");
-			},
-		);
-		await new Promise((resolve, reject) => {
-			const server = app.listen(this.options.port, LISTEN_ADDRESS, () => {
-				this.server = server;
-				const address = server.address();
-				if (typeof address === "object" && address) {
-					this.boundPort = address.port;
-				} else {
-					this.boundPort = this.options.port;
-				}
-				this.options.onLog?.(
-					`Proxy Server running on http://${LISTEN_ADDRESS}:${this.boundPort}`,
-				);
-				resolve();
-			});
-			server.once("error", reject);
-		});
-	}
-	async stop() {
-		if (!this.server) {
-			return;
-		}
-		const server = this.server;
-		this.server = null;
-		this.boundPort = null;
-		await new Promise((resolve, reject) => {
-			server.close((error) => {
-				if (error) {
-					reject(error);
-					return;
-				}
-				resolve();
-			});
-		});
-	}
-	async handleCompletionRequest(req, res, routeKind) {
-		const body = req.body;
-		const modelReference = this.resolveModelReference(
-			body.model,
-			req.params.modelId,
-		);
-		if (!modelReference) {
-			res.status(400).json({ error: "Model ID is required" });
-			return;
-		}
-		const model = this.resolveModel(modelReference);
-		if (!model) {
-			res.status(404).json({ error: `Model ${modelReference} not found` });
-			return;
-		}
-		const apiKey = await this.options.modelRegistry.getApiKey(model);
-		if (!apiKey) {
-			res
-				.status(401)
-				.json({ error: `No credentials for provider ${model.provider}` });
-			return;
-		}
-		const abortController = new AbortController();
-		req.once("close", () => abortController.abort());
-		const maxTokens =
-			routeKind === "openai"
-				? body.max_tokens
-				: body.generationConfig?.maxOutputTokens;
-		const context = this.normalizeContext(body, routeKind);
-		const options = {
-			apiKey,
-			temperature: body.temperature,
-			maxTokens,
-			signal: abortController.signal,
-		};
-		const eventStream = this.streamModel(model, context, options);
-		const shouldStream =
-			routeKind === "google" ? body.stream !== false : body.stream === true;
-		if (shouldStream) {
-			await this.sendStreamingResponse(eventStream, res, routeKind, model);
-			return;
-		}
-		await this.sendBufferedResponse(eventStream, res, routeKind, model);
-	}
-	resolveModelReference(bodyModel, pathModelId) {
-		return bodyModel ?? pathModelId;
-	}
-	resolveModel(modelReference) {
-		const normalizedReference = modelReference.toLowerCase();
-		const exact = this.options.modelRegistry
-			.getAll()
-			.find(
-				(model) =>
-					`${model.provider}/${model.id}`.toLowerCase() ===
-						normalizedReference ||
-					model.id.toLowerCase() === normalizedReference,
-			);
-		if (exact) {
-			return exact;
-		}
-		const slashIndex = modelReference.indexOf("/");
-		if (slashIndex === -1) {
-			return undefined;
-		}
-		const provider = modelReference.slice(0, slashIndex);
-		const modelId = modelReference.slice(slashIndex + 1);
-		return this.options.modelRegistry.find(provider, modelId);
-	}
-	normalizeContext(body, routeKind) {
-		if (routeKind === "google") {
-			return this.normalizeGoogleContext(body);
-		}
-		return this.normalizeOpenAiContext(body);
-	}
-	normalizeOpenAiContext(body) {
-		const messages = body.messages ?? [];
-		const systemPrompt = messages.find(
-			(message) => message.role === "system",
-		)?.content;
-		const normalizedMessages = messages
-			.filter((message) => message.role !== "system")
-			.map((message) => this.normalizeOpenAiMessage(message));
-		return {
-			systemPrompt: typeof systemPrompt === "string" ? systemPrompt : undefined,
-			messages: normalizedMessages,
-		};
-	}
-	normalizeGoogleContext(body) {
-		const systemPrompt =
-			body.systemInstruction?.parts?.map((part) => part.text ?? "").join("") ||
-			undefined;
-		const normalizedMessages = (body.contents ?? [])
-			.map((content) => {
-				const textContent = (content.parts ?? [])
-					.filter((part) => typeof part.text === "string")
-					.map((part) => ({ type: "text", text: part.text ?? "" }));
-				if (content.role === "user") {
-					return this.createUserMessage(textContent);
-				}
-				return this.createAssistantMessage(textContent);
-			})
-			.filter((message) => message.content.length > 0);
-		return {
-			systemPrompt,
-			messages: normalizedMessages,
-		};
-	}
-	normalizeOpenAiMessage(message) {
-		if (message.role === "assistant") {
-			return this.createAssistantMessage(
-				this.normalizeContent(message.content),
-			);
-		}
-		return this.createUserMessage(this.normalizeContent(message.content));
-	}
-	createUserMessage(content) {
-		return {
-			role: "user",
-			content,
-			timestamp: Date.now(),
-		};
-	}
-	createAssistantMessage(content) {
-		const normalizedContent =
-			typeof content === "string" ? [{ type: "text", text: content }] : content;
-		return {
-			role: "assistant",
-			content: normalizedContent,
-			api: "google-gemini-cli",
-			provider: "google-gemini-cli",
-			model: "proxy",
-			usage: {
-				input: 0,
-				output: 0,
-				cacheRead: 0,
-				cacheWrite: 0,
-				totalTokens: 0,
-				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
-			},
-			stopReason: "stop",
-			timestamp: Date.now(),
-		};
-	}
-	normalizeContent(content) {
-		if (typeof content === "string") {
-			return content;
-		}
-		return (content ?? [])
-			.filter((part) => typeof part.text === "string")
-			.map((part) => ({ type: "text", text: part.text ?? "" }));
-	}
-	async sendStreamingResponse(eventStream, res, routeKind, model) {
-		res.status(200);
-		res.setHeader(
-			"Content-Type",
-			routeKind === "openai" ? SSE_CONTENT_TYPE : NDJSON_CONTENT_TYPE,
-		);
-		res.setHeader("Cache-Control", "no-cache");
-		res.setHeader("Connection", "keep-alive");
-		for await (const event of eventStream) {
-			if (event.type === "text_delta") {
-				if (routeKind === "openai") {
-					res.write(
-						`data: ${JSON.stringify(this.buildOpenAiChunk(model, event.delta))}\n\n`,
-					);
-				} else {
-					res.write(`${JSON.stringify(this.buildGoogleChunk(event.delta))}\n`);
-				}
-			}
-			if (event.type === "done") {
-				if (routeKind === "openai") {
-					res.write("data: [DONE]\n\n");
-				}
-				res.end();
-				return;
-			}
-			if (event.type === "error") {
-				if (!res.headersSent) {
-					res
-						.status(500)
-						.json({ error: event.error.errorMessage ?? "Proxy stream failed" });
-				} else {
-					res.end();
-				}
-				return;
-			}
-		}
-		res.end();
-	}
-	async sendBufferedResponse(eventStream, res, routeKind, model) {
-		const assistantMessage = await eventStream.result();
-		const text = this.extractText(assistantMessage);
-		if (routeKind === "openai") {
-			res.json({
-				id: `chatcmpl-${Date.now()}`,
-				object: "chat.completion",
-				created: Math.floor(Date.now() / 1000),
-				model: model.id,
-				choices: [
-					{
-						index: 0,
-						message: { role: "assistant", content: text },
-						finish_reason: "stop",
-					},
-				],
-				usage: assistantMessage.usage,
-			});
-			return;
-		}
-		res.json({
-			candidates: [
-				{
-					content: {
-						parts: [{ text }],
-					},
-				},
-			],
-			usageMetadata: assistantMessage.usage,
-		});
-	}
-	extractText(message) {
-		return message.content
-			.filter((content) => content.type === "text")
-			.map((content) => content.text)
-			.join("");
-	}
-	buildOpenAiChunk(model, delta) {
-		return {
-			id: `chatcmpl-${Date.now()}`,
-			object: "chat.completion.chunk",
-			created: Math.floor(Date.now() / 1000),
-			model: model.id,
-			choices: [{ index: 0, delta: { content: delta }, finish_reason: null }],
-		};
-	}
-	buildGoogleChunk(delta) {
-		return {
-			candidates: [
-				{
-					content: {
-						parts: [{ text: delta }],
-					},
-				},
-			],
-		};
-	}
-}
-export function createProxyServer(options) {
-	return new ProxyServer(options);
-}
diff --git a/src/resources/extensions/sf/skills/loader.js b/src/resources/extensions/sf/skills/loader.js
index 57e1882bf..b2eebd593 100644
--- a/src/resources/extensions/sf/skills/loader.js
+++ b/src/resources/extensions/sf/skills/loader.js
@@ -17,9 +17,14 @@ import {
  * Load all valid skills from all sources.
  *
  * Returns array of skill records with validation errors attached.
+ * Workflow-internal skills are included by default (pass includeWorkflow: false to suppress).
  */
 export function loadSkills(projectPath, options = {}) {
-	const discovered = discoverAllSkills(projectPath, options);
+	const resolvedOptions = {
+		includeWorkflow: true,
+		...options,
+	};
+	const discovered = discoverAllSkills(projectPath, resolvedOptions);
 	const skills = [];
 
 	for (const { name, path, source } of discovered) {
diff --git a/src/resources/extensions/sf/tests/skills.test.mjs b/src/resources/extensions/sf/tests/skills.test.mjs
index b71a69ce5..d10092303 100644
--- a/src/resources/extensions/sf/tests/skills.test.mjs
+++ b/src/resources/extensions/sf/tests/skills.test.mjs
@@ -204,7 +204,7 @@ describe("skill loading", () => {
 		createSkill("skill-a");
 		createSkill("skill-b", { permissionProfile: "trusted" });
 
-		const skills = loadSkills(tmpDir);
+		const skills = loadSkills(tmpDir, { includeWorkflow: false });
 		expect(skills).toHaveLength(2);
 		expect(skills.every((s) => s.valid)).toBe(true);
 		expect(skills.some((s) => s.name === "skill-a")).toBe(true);
@@ -217,7 +217,7 @@ describe("skill loading", () => {
 		mkdirSync(badDir, { recursive: true });
 		writeFileSync(join(badDir, "SKILL.md"), "No frontmatter here.");
 
-		const skills = loadSkills(tmpDir);
+		const skills = loadSkills(tmpDir, { includeWorkflow: false });
 		expect(skills).toHaveLength(2);
 		const bad = skills.find((s) => s.name === "bad-skill");
 		expect(bad).toBeTruthy();
@@ -229,7 +229,7 @@ describe("skill loading", () => {
 		createSkill("normal-skill", { permissionProfile: "normal" });
 		createSkill("trusted-skill", { permissionProfile: "trusted" });
 
-		const skills = loadSkills(tmpDir);
+		const skills = loadSkills(tmpDir, { includeWorkflow: false });
 		const permitted = getPermittedSkills(skills, "normal");
 		expect(permitted).toHaveLength(2);
 		expect(permitted.some((s) => s.name === "restricted-skill")).toBe(true);
@@ -243,7 +243,7 @@ describe("skill loading", () => {
 		createSkill("universal-skill", { triggers: ["*"], modelInvocable: true });
 		createSkill("user-only", { triggers: ["*"], modelInvocable: false });
 
-		const skills = loadSkills(tmpDir);
+		const skills = loadSkills(tmpDir, { includeWorkflow: false });
 		const buildSkills = getModelInvocableSkills(skills, "build");
 		expect(buildSkills).toHaveLength(2);
 		expect(buildSkills.some((s) => s.name === "build-skill")).toBe(true);