VrianCao
179 posts
80 hours of AI pair programming. Here's what context-mode saved me.
→ $487.20 in API costs. Opus pricing. Real money, not estimates.
→ 22.3 hours of re-explaining context after compaction. Time I got back to ship.
→ 268 sessions resumed from memory. The agent never asked "what were we doing?"
→ 47 preferences auto-learned. "use TS strict" once, remembered forever.
→ 14,847 events indexed. Searchable across every session, every project.
Without context-mode |████████████████████████████████████████| 6.2 MB
With context-mode |█░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░| 124 KB
98% of raw data never entered my conversation.
That's a 50× longer session. Same context window.
Multiply across a 13-engineer team:
$487 × 13 = $6,331/month saved
22 hours × 13 = 286 hours/month recovered
Open source. Local-first. No telemetry. No account. No SaaS lock-in.
github.com/mksglu/context…
English
@mksglu Done it
github.com/mksglu/context…
(Woo a pretty cool PR Number
English
@VrianCao oh so sorry, it's my fault. Will fix that. Can you open a issue or PR, please.
English
@mksglu Oh, I see. That’s because ctx-insight uses an environment variable INSIGHT_CONTENT_DIR which defaults to the Claude code directory. When using pi (or others), you’d expect the agent to set the correct directory manually, but the MCP doesn't have an argument to fill that in.
English
@VrianCao It's seems a bug i think. Let's investigate it as deep-dive. When you have a time please cross-check with `ctx-stats` and `ctx-insight` Thanks!
English
@VrianCao You can run `ctx-insight` for global stat.
English
116.5K+ users · 11.7K+ GitHub stars · 793 forks · 14 platforms
Thanks a lot for sharing, really appreciate it !🫡
中国开发者社区的朋友们，期待你们的支持与反馈。
我非常重视你们的意见，也感谢一直以来的支持。当前有一个关于代理支持的 PR，但我不确定是否真的有必要或实现是否正确。如果这是中国网络环境中的实际需求，请多多指点。
感谢大家的帮助。
github.com/mksglu/context…
GitHubDaily@GitHub_Daily
最近看到一个开源项目 Context Mode，能有效解决 AI 编程工具上下文被超出的问题。 核心思路是让原始数据留在沙盒里，只把处理结果送进上下文窗口。 据介绍，能把 315 KB 的原始输出压缩到 5.4 KB，长度节省高达 98%。 同时用本地数据库记录会话状态，对话压缩后也能无缝恢复。 GitHub：github.com/mksglu/context… 支持 14 个平台，包括 Claude Code、Cursor、Gemini CLI、VS Code Copilot 等主流 AI 编程工具。 内置 11 种语言运行时，还有知识库索引和智能搜索功能，方便按需检索而不是一股脑塞进上下文。 所有数据都在本地处理，不联网不上传。如果你经常遇到 AI 编程到一半就出现幻觉的情况，不妨试试这个工具。
中文
@badlogicgames Pi is definitely the best agent core. Love it so much.!
English
ok, i already posted this but holy shit it's built on pi?!
#L3" target="_blank" rel="nofollow noopener">github.com/withastro/flue…
this makes me super happy!
fks@FredKSchott
Introducing Flue — The First Agent Harness Framework Flue is a TypeScript framework for building the next generation of agents, designed around a built-in agent harness. Flue is like Claude Code, but 100% headless and programmable. There's no baked in assumption like requiring a human operator to function. No TUI. No GUI. Just TypeScript. But using Flue feels like using Claude Code. The agents you build act autonomously to solve problems and complete tasks. They require very little code to run. Most of the "logic" lives in Markdown: skills and context and AGENTS.md. Flue is like Astro or Next.js for agents (not surprising, given my background 🙃). It's not another AI SDK. It's a proper runtime-agnostic framework. Write once, build, and deploy your agents anywhere (Node.js, Cloudflare, GitHub Actions, GitLab CI/CD, etc). We originally built Flue to power AI workflows inside of the Astro GitHub repo. But then @_bgiori got his hands on it, and we realized that every agent needs a framework like Flue, not just us. Check it out! It's early, but I'm curious to hear what people think. Are agents ready for their library -> framework moment?
English
@OpenAI haha I'm right
See u later 🫡
x.com/i/status/20465…
VrianCao@VrianCao
My GPT-5.4 Pro just froze mid-thought. Even after I steered it to 'Continue,' it only thought for a second before getting stuck again. Considering yesterday’s massive OpenAI outage, I can’t think of any other reason besides some major infrastructure adjustments. Big model smell.
English
My GPT-5.4 Pro just froze mid-thought. Even after I steered it to 'Continue,' it only thought for a second before getting stuck again. Considering yesterday’s massive OpenAI outage, I can’t think of any other reason besides some major infrastructure adjustments. Big model smell.
English
@iamai_omni @dongxi_nlp @sama OAI的Intra比Anthropic好到不知道哪里去了
自己看看可用性检测和Status吧，都不是一个量级的东西
但凡了解过一些的都不会喷OAI的Intra
中文
Impressive work from Kimi team!
Kimi.ai@Kimi_Moonshot
Meet Kimi K2.6: Advancing Open-Source Coding 🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2) What's new: 🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization). 🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D. 🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files. 🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops. 🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop. - K2.6 is now live on kimi.com in chat mode and agent mode. For production-grade coding, pair K2.6 with Kimi Code: kimi.com/code - 🔗 API: platform.moonshot.ai 🔗 Tech blog: kimi.com/blog/kimi-k2-6 🔗 Weights & code: huggingface.co/moonshotai/Kim…
English
@michellechen Spotted your name! Maybe we’ll see your name in the blog post again when GLM-5.1 launches on Cloudflare LOL. Anyway, great job—not just on these model inference services, but the entire Agent Week. It’s been absolutely impressive!
English
serving models is hard. serving extra large language models with good quality, throughput, reliability, pricing, and gpu utilization is really hard.
a look behind the scenes of how we do it on workers ai this agents week. there’s always more to do, but we’re just getting started
blog.cloudflare.com/high-performan…
English
