Recent Posts
Skill Recorder: Record a Task Once, Let Your AI Agent Repeat It Forever
1 months ago 高效码农Microsoft’s Skill Recorder: Record a Task Once, Let Your AI Agent Repeat It Do you ever find yourself doing the same sequence of clicks, forms, and file uploads every week? You know the steps by heart, but automating them feels like more work than just doing them manually again. Microsoft recently open‑sourced a tool called Skill Recorder that tackles this from a different angle. Instead of giving you another screen recorder or a fragile macro tool, Skill Recorder captures your screen and actions while you work, then uses GitHub Copilot to turn that recording into a structured procedure. That procedure …
LangChain OpenRouter: How to Access 400+ LLMs Through One API Without Rewriting Chains
1 months ago 高效码农LangChain + OpenRouter: A Practical Guide to Calling 400+ LLMs Through One API How to integrate OpenRouter’s unified model gateway into your LangChain applications without rewriting your chains. If you’re building AI apps with LangChain, you’ve probably faced the provider dilemma. Lock into OpenAI’s GPT family? Bet on Anthropic’s Claude? Keep a back door open for Google’s Gemini? Every choice means a separate API key, a different parameter schema, and its own failure modes. Worse, when you want to A/B test GPT-4 against Claude, or give your users in China a DeepSeek fallback, nearly every line of your chain code …
Codex DeepSeek Setup Errors: 5 Common Failures and How to Fix Them
1 months ago 高效码农Codex + DeepSeek Setup Failing? Here’s Every Error We Hit (And How to Fix Each One) If you’re routing Codex through CC Switch to run DeepSeek instead of GPT, you’ll likely run into five specific failures, in this order: a curl syntax mistake, a local proxy network mismatch, a provider config gap, an upstream protocol mismatch, and a malformed test request. This guide walks through each one, in the order they actually show up, with the reasoning behind each fix so you can diagnose new variants yourself. Why DeepSeek Doesn’t Just Work With Codex Out of the Box Codex talks …
ARC-AGI-3 Harness Fail: Why GPT-5.6 Sol Scored Only 7.8% – And the Simple Fix That Tripled Its Score
1 months ago 高效码农GPT-5.6 Sol Scored 7.8% on ARC-AGI-3. The Problem Was the Harness, Not the Model. When we first pulled up the ARC-AGI-3 leaderboard, our immediate reaction was confusion. GPT-5.6 Sol had already solved the cycle double cover conjecture, a longstanding open problem in mathematics. It beat Pokémon FireRed using a vision-only setup. Yet on ARC-AGI-3, a benchmark built from 2D puzzle games, it managed a mere 7.8%. GPT-5.5’s performance was even more extreme, scoring 0.4%, which essentially means it failed to play the games at all. How does a model capable of beating Pokémon get stuck on simple 2D puzzles? The …
WeChat Decryption: How to Export Your Encrypted macOS Chat History Using wechat-exporter
1 months ago 高效码农My WeChat Chat History, Finally Freely Queryable A lot of people don’t realize this, but the chat history we generate every day on WeChat actually sits on our computer’s hard drive—encrypted. Want to back it up? Export it? Analyze it? You’re stuck without the key that unlocks it. That’s exactly what wechat-exporter solves. It does one thing directly: it takes your macOS WeChat local chat history and decrypts the entire thing into plain SQLite databases. After that, you can query, export, or analyze however you like. This post walks through what it does, how to use it, and what’s happening …
Claude Code Opens Bun Help Screen? Step-by-Step Fix for Windows & npm
1 months ago 高效码农Claude Code CLI Opens the Bun Help Screen Instead of Starting? A Practical Troubleshooting Guide for PATH, npm, and Installation Issues Running claude Shows Bun Instead of Claude Code If you type: claude and instead of launching Claude Code you see something like this: Bun is a fast JavaScript runtime, package manager, bundler, and test runner. Usage: bun <command> … you are not looking at a Claude Code error. What actually happened is that your command execution chain ended up invoking Bun, not Claude Code. This situation can be confusing because the command itself exists and executes successfully. There is …
Excel AI Automation: Why File-Parsing Fails and How Native COM via ExcelMcp Empowers AI Agents
1 months ago 高效码农How to Drive Native Excel with ExcelMcp for AI Agent Automation How can you enable AI assistants to directly interact with Microsoft Excel to build complex data models, generate PivotTables, and refresh Power Query connections? Traditional automation approaches rely on third-party libraries to parse raw .xlsx files. This strategy strips away dynamic formula recalculations and prevents you from triggering live VBA macros or internal Excel calculation engines. ExcelMcp bridges this gap using the Model Context Protocol (MCP) and Windows COM automation, giving AI tools like GitHub Copilot, Claude, and Cursor direct control over a running Microsoft Excel instance. What ExcelMcp …
Pi Coding Agent No Matching Models? Fix Alibaba Token Plan Issue
1 months ago 高效码农Pi Coding Agent CLI + Alibaba Cloud Model Studio Token Plan: Troubleshooting “No Matching Models” After Successful Login Problem Overview: Why Does Pi Coding Agent Show No Models After Alibaba Token Plan Login? Pi Coding Agent CLI can successfully authenticate with Alibaba Cloud Model Studio Token Plan, but still fail to display any available models. The typical symptom looks like this: Only showing models from configured providers. Use /login to add providers. No matching models At first glance, this error looks like an authentication problem. However, during troubleshooting, the actual issue was different. The login process was successful: ❀ Alibaba …
DashScope Pi Agent Configuration: Step-by-Step Guide to Token Plan Setup
1 months ago 高效码农How to Configure DashScope Token Plan in Pi Agent To connect Alibaba Cloud DashScope’s Token Plan to Pi Agent, you need to register [https://coding.dashscope.aliyuncs.com/v1](https://coding.dashscope.aliyuncs.com/v1) as an OpenAI-compatible service within Pi’s local configuration files. By default, Pi’s interactive terminal login menu (/login) only lists natively pre-configured providers. You won’t see an option to enter a custom API endpoint directly in that UI menu. The solution is straightforward: define your service configuration in a local models.json file. Once the base URL, authentication key, and model IDs are mapped correctly, restart Pi and select your custom model through the /model menu. Here is …
Claude Code Session Recovery: How to Instantly Resume After Ctrl+C in PyCharm
1 months ago 高效码农Gemini API Managed Agents Just Got Free Tier, Budget Controls, and Cron Triggers – Here’s What That Means for You
1 months ago 高效码农Gemini API Managed Agents Just Got Free Tier, Budget Controls, and Cron Triggers – Here’s What That Means for You If you’ve been following the Gemini API’s managed agents (the ones that run in an isolated sandbox with code execution, file management, and network access), you probably had two reactions: “this is incredibly useful” and “how do I keep costs under control?” Google just addressed both. The latest updates – free tier availability, budget guardrails, and scheduled triggers – turn managed agents from a powerful but slightly risky tool into something you can actually deploy without holding your breath. I’ve …
CoPaw JSONDecodeError: How a Corrupted Session State File Crashed Our AI (And How We Fixed It)
1 months ago 高效码农How We Diagnosed and Fixed a CoPaw 1.0.2 JSONDecodeError Caused by a Corrupted Session State File Keywords: CoPaw 1.0.2, JSONDecodeError, CoPaw session state, AgentScope Runtime, Python JSON error, CoPaw troubleshooting, corrupted JSON file, FastAPI 500 error, AI agent debugging Introduction While running CoPaw 1.0.2, we encountered a recurring error that prevented chat sessions and agent tasks from executing correctly. The application started normally, models loaded successfully, but any attempt to access chat history or execute an agent workflow resulted in a failure. The root cause turned out to be a surprisingly common issue in stateful AI applications: an empty or …
GPT-5.6 Prompt Optimization: The Lean Way to 66% Token Cuts
1 months ago 高效码农Stop Over-Prompting GPT-5.6: Why Leaner Instructions Deliver Better Results Working with GPT-5.6 is like onboarding a brilliant but impatient new hire. The more you try to micromanage every step, the more likely it is to go off the rails. Instead of spending hours writing lengthy system prompts, focus your energy on clarifying the outcome and defining the stopping conditions. Earlier this year, OpenAI rolled out GPT-5.6. Our team immediately migrated our internal coding-agent pipeline to test it out. The results were counterintuitive: our old, verbose prompts—the ones that worked perfectly fine on GPT-5.5—actually hurt performance. When we stripped down the …
Unified Multimodal Generation: One Model for All Vision Tasks? SenseNova‑Vision Answers
1 months ago 高效码农One Model for All Vision Tasks? SenseNova‑Vision Answers with Unified Multimodal Generation Computer vision has long been a patchwork of isolated islands. Object detection spits out bounding boxes, segmentation produces masks, depth estimation gives grayscale maps, and 3D reconstruction deals with point clouds and camera parameters. Each task has its own output format, its own network heads, its own evaluation metrics. Trying to do multiple things with a single model usually means wrestling with incompatible output spaces. SenseNova‑Vision tackles a straightforward question: can we squeeze the entire spectrum of computer vision tasks into a single generative framework? Much like GPT …
How to Extract Real Research Innovation from Literature Gaps: The Evidence Problem Card Method
1 months ago 高效码农How RW Research Skill Extracts Actual Research Innovation from Literature Gaps A literature gap is just a hole in the data. It is not a research plan. Many researchers hit a wall after reading dozens of papers and building massive extraction tables. They stare at the screen, unable to decide what to actually study. You spot an author writing “further research is needed” in the discussion section. You notice a population no one has touched, a method rarely applied in a specific niche, or two papers arriving at opposite conclusions. All of these are gaps. But a gap merely indicates …
Gemini 3.6 Flash Migration: 5 Critical API Changes That Will Break Your Code (And How to Fix Them)
1 months ago 高效码农Gemini 3.6 Flash & 3.5 Flash-Lite Are GA: What Breaks, What’s Better, and How to Migrate Google just made Gemini 3.6 Flash and Gemini 3.5 Flash-Lite generally available. That means they are production-ready, right now. I’ve dug through the official release notes, and here’s the thing: these aren’t just “better models.” They come with API changes that will break your existing code if you aren’t paying attention. Specifically, temperature, top_p, and top_k are being shown the door. This post cuts through the marketing. I’ll tell you exactly what each model does well, which one you should pick for your workload, …
Financial Agent Skills: The Playbook for Building AI Digital Employees
1 months ago 高效码农Financial Industry Agent 100 Skills: From General AI to Digital Employees – A Practical Guide By 2026, AI adoption in China’s financial sector has crystallized into concrete numbers: 7‑day iteration cycles, over 2,500 deployed agents, and end‑to‑end process re‑engineering. Large models plus agentic systems have proven the arrival of what some call the “Loopy Era” – agents that can autonomously design experiments, refine models, and iterate without human supervision. This isn’t a roadmap anymore; it’s what’s happening on the ground. Yet one question keeps coming up in every bank, insurer, and asset manager I talk to: How do you make …
CodeX Reconnecting 5/5? Fix WebSocket Issues with This Simple HTTP Config Twea
1 months ago 高效码农CodeX Stuck on “Reconnecting 5/5”? Stop Blaming the Model—Here’s How to Skip WebSocket and Go Straight to HTTP You ask CodeX to change one line of code. Instead of editing, it spends 30 seconds showing “Reconnecting 1/5 … 5/5”, then “Thinking…” for a while longer. You stare at the terminal: is it deep in reasoning, or just stalling? It’s neither. The model hasn’t even started. The network layer is busy retrying a path that won’t work—five times, back to back—before it finally falls back to a working connection. Your internet speed isn’t the problem. The culprit is your proxy, …
Production-Grade AI Agent Architecture: The Hermes Runtime That Keeps Agents Alive
1 months ago 高效码农Production-Grade AI Agent Architecture: A Deep Dive into the Hermes Runtime Treating an AI Agent as nothing more than a ReAct loop wrapped around an LLM works fine in a demo. Ship it to production, and it’ll collapse within three days. It’s not the model’s fault—it’s the missing runtime. The Real Question: What Is an Agent Framework Actually Solving? Over the past two years, I’ve built my fair share of Agent applications. I started with quick LangChain prototypes—basic chatbots that barely held context. But eventually, I needed something that could run inside Feishu (Lark), read local files, call internal APIs, …
Claude Loops: Stop Using AI as a One-Shot Calculator and Build Reusable Workflows
1 months ago 高效码农Stop Using Claude as a One-Shot Calculator: Building Reusable Workflows with Claude Loops Most people still interact with Claude in a one-shot pattern. You write a prompt. Claude responds. The session ends. The next day, you rebuild the exact same context from scratch. That works fine for isolated tasks. But for repetitive work—daily code reviews, CI triage, project monitoring, meeting follow-ups—rebuilding context daily isn’t a workflow. It’s manual labor with AI sprinkles on top. The better abstraction is a loop. What a Claude Loop Actually Is (And Why It’s Not Just “Asking Again”) A Claude Loop is not simply “asking …