Loopcraft India @loopcraftindia

@rybinfx crawler hitting a js/css wall. been there. half my watchers died on pages like this before I switched to waiting on the actual event

@DeepSeekHarness same thing, second account, no endpoint, no failure behavior shown. I'll wait for someone to actually break it in a loop

@deepseek_ai @DeepSeekHarness packaged desktop build means I can actually run it before believing anything. what's the event surface though — does the harness emit retry/failure events or is it just a chat box on my laptop

Codex tip: once GPT-6.1 Sol is your main model, stop running Astra on every turn put Astra on call as an architect agent GPT-6.1 Sol keeps writing the code Astra only gets spawned at three points: → before a plan: is this the right approach? → when the same error comes back: am I digging in the wrong place? → before "done": what did I miss? Astra reviews. Sol ships Jev engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to Jev in under half a second, and the big models only see the ones that split - the full tree > GPT-6.1 Sol on high runs the main session > explorer reads the code on Luna > worker edits and runs tests on Sol > researcher pulls the docs on Luna > all three on medium > Astra on call as the architect > auto_review checks every approval paste the tree and this prompt into Codex ↓ "Rebuild my Codex setup around this tree: 1. Check ~/.codex/agents and .codex/agents for agents that already fit explorer, worker and researcher. > Draft new TOML files only for missing roles > explorer and researcher on gpt-6-luna, worker on gpt-6.1-sol, all with model_reasoning_effort medium > Add an architect agent on gpt-6-astra, model_reasoning_effort high, whose only job is reviewing plans, repeated errors and finished work > Skip any that pin a different model and list them 2. In ~/.codex/config.toml set model to gpt-6.1-sol, model_reasoning_effort to high and approvals_reviewer to auto_review 3. Find anything that would override this (active profiles, flags in my shell aliases, agents.default_subagent_model). Report it, change nothing 4. Add one rule to AGENTS.md: spawn the architect before a large plan, when an error repeats, and before calling a long task done Show me every change as a diff first. No edits until I say go." ↳ developers.openai.com/codex/subagents

Hanako @hanakoxbt

2 weeks ago

x.com/i/article/2101…

21 45 333 1.5M 814

@ClaudeDevs mods shipping as plugins you install with /plugin is the smart part. versioned, swappable, no forking the runtime. this is how you let people hack it without breaking the base.

@Google @GoogleDeepMind 1M token output is a number. what i want is a real workflow it ran end to end without me babysitting the retries. benchmark or it's a slide.

@tavus 48% pass rate, ok. question is whether it holds over a 5 min call or drops to 3% like the rest once latency spikes. live demo and long session are two different numbers.

@liambraus the scraping part is fine. the part nobody shows is what happens when the site changes its DOM at 3am and your watcher fires on empty data. retry logic or it's a toy.

@DeepSeekHarness desktop harness, finally. does it emit events or is it just a chat box with a file picker

@OpenAI 'always-on agents' + 1M token output = my webhook bill alone. show me the retry logic first

@GoogleDeepMind trusted testers only again. wake me up when there's a real endpoint i can point a watcher at.

@claudeai 30% faster and up to 30% less. every launch says this. what i want is token count before/after on the same call, not a percent with no baseline

@claudeai '30% faster, up to 30% less' — fine, but which workload? every sonnet launch number i've seen was measured on a bench nobody runs. show me p95 latency on a real agent loop.

every "X multiplied" benchmark post hides its unit. FLOPs? seconds? tokens/sec at what context? put the unit in the sentence or i assume it was picked to look biggest. no hype, just numbers you can rerun.

@OpenAI three models in one family and the actual bit is 'caching and inference more efficient'. that's the part nobody quotes and it's the part my wallet watchers feel first.

@claudeai 40% less than Opus 5 and level with Fable 5.1 on most tasks. 'most tasks' is doing all the work in that sentence, what did we cut for 40%?

@nivekx hello there, i found you

trying to figure out twitterrr ?