Chris Cook zirkelc

<core_identity> You are an assistant called Cluely, developed and created by Cluely, whose sole purpose is to analyze and solve problems asked by the user or shown on the screen. Your responses must be specific, accurate, and actionable. </core_identity>

<general_guidelines>

NEVER use meta-phrases (e.g., "let me help you", "I can see that").
NEVER summarize unless explicitly requested.
NEVER provide unsolicited advice.
NEVER refer to "screenshot" or "image" - refer to it as "the screen" if needed.
ALWAYS be specific, detailed, and accurate.

GitHub Release Automation Tools

This is a list of tools to automate GitHub releases including the generation of changelogs, version bumps and publishing of releases or packages.

	#!/usr/bin/env bash
	#
	# CloudWatch Logs Insights freshness repro.
	#
	# Writes a log event, then asks Logs Insights for it every few seconds with four
	# queries that differ only in shape, and shows live which ones find it:
	#
	# nofilter no filter at all (the event is matched on the client)
	# filter filter @message like /<marker>/
	# parse parse @message ... \| filter @message like /<marker>/ and ispresent(level)

 /**
  * Azure OpenAI rejects some GPT-5.4 prompts above 272k tokens with
  * `context_length_exceeded`, although the model has a 1M context window.
  *
  * Which prompts fail depends on how many characters they have per token, not on
  * their token count: English at 285k tokens passes, Japanese, Chinese or dense
  * ASCII at 285k tokens fails, and Japanese at 700k tokens passes again. The same prompts
  * sent to OpenAI directly work.
  *
  * This script sends synthetic prompts of one exact token size but different text

 #!/usr/bin/env node
 /**
  * Azure OpenAI rejects some GPT-5.4 prompts above 272k tokens with
  * `context_length_exceeded`, although the model has a 1M context window.
  *
  * Which prompts fail depends on how many characters they have per token, not on
  * their token count: English at 285k tokens passes, Japanese, Chinese or dense
  * ASCII at 285k tokens fails, and Japanese at 700k tokens passes again. The same prompts
  * sent to OpenAI directly work.
  *

 #!/usr/bin/env node
 /**
  * Azure OpenAI rejects some GPT-5.4 prompts above 272k tokens with
  * `context_length_exceeded`, although the model has a 1M context window.
  *
  * Which prompts fail depends on how many characters they have per token, not on
  * their token count: English at 285k tokens passes, Japanese or dense ASCII at
  * 285k tokens fails, and Japanese at 700k tokens passes again. The same prompts
  * sent to OpenAI directly work.
  *

	import "dotenv/config";
	import { openai } from "@ai-sdk/openai";
	import { type InferUITools, stepCountIs, streamText, tool, type UIMessage } from "ai";
	import { z } from "zod";
	import { pipe, toolCall } from "ai-stream-utils";

	const tools = {
	weather: tool({
	description: `Get the current weather for a location`,
	inputSchema: z.object({

	import { UIMessage, InferUITools, tool } from 'ai';

	type ToolSet = ReturnType<typeof createToolSet>['tools'];

	// Create a tool set
	// Optionally accept common tool params like a UIStreamWriter
	export const createToolSet = () => {
	const tools = {
	firstTool: tool({
	description: 'Greets the user',

	import type { LanguageModelV2 } from '@ai-sdk/provider';
	import { APICallError } from 'ai';
	import type { Retryable, RetryModel } from '../create-retryable-model.js';
	import { isErrorAttempt, isResultAttempt } from '../create-retryable-model.js';
	import { isObject, isString } from '../utils.js';

	/**
	* Fallback to a different model if Anthropic returns an "overloaded" error.
	*
	* ```

	import fs from 'node:fs/promises';
	import path from 'node:path';

	async function* readAllFiles(dir: string): AsyncGenerator<string> {
	// use fs.readdirSync() to avoid async
	const files = await fs.readdir(dir, { withFileTypes: true });

	for (const file of files) {
	if (file.isDirectory()) {
	yield* readAllFiles(path.join(dir, file.name));