vianel (@vianeltxt) header
vianel (@vianeltxt) profile picture

i'm one with the force and the force is with me. he/him

Nvim
Joined February 2010
Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)Photo by vianel (@vianeltxt)
Introducing Gemini Decision 1, our new frontier decision model in the Gemini API. It is $0.10 per 1M input tokens with no additional cost. With support for image, audio, video, and PDF natively, 256K context window, & 1K questions. You can try it for free in the API!
595
632
358
10,631
798,540
vianel retweeted
LLMs are still SO bad at building tools for agents. Every time I try to make Opus set up an MCP authentication path, it makes horrible assumptions and builds something that straight up doesn't work
254
38
43
2,155
169,385
2010: -write code -ship it -move on 2025: -write code -AI writes better code -spend the day convincing yourself the AI version is wrong -ship the AI version -it breaks in production anyway
4
8
1
97
13,456
What's better than refining prompts by hand? Letting your own decisions do it. I built wonder, an open source CLI that turns a GitHub issue into a reviewed pull request using the coding agents you already have Claude Code, Codex and GitHub Copilot It runs like a small factory 🧵
6
4
165
What did this issue cost me? Every agent call is metered: tokens, dollars and agent time. You get a line like ✓ merged PR #16 · spent $0.42 in 2 min of agent time I'm looking for the first users. If you have a repo with a backlog of small issues and one of these CLIs installed,
1
17
Felix Rieseberg, an engineering manager at Anthropic, says the coding interview is dead. Claude now beats candidates at the React challenges and algorithm puzzles we’ve used to screen engineers for years. I agree, and I think the reason matters.
1
42
So my suggestion for interviewers: give candidates a real problem, let them use their agents, and watch how they think. Look at what they delegate, what they question, and which trade-offs they catch before the agent does
1
22