598 Photos and videos
I spent the past week building a marketing automation harness as an OpenClaw plugin. It runs in production now, publishes real posts to a real client page, with an LLM in the loop.
My only rule from day 1: the model never touches anything irreversible.
some field notes 🧵
2
1
8
129
🦞 OpenClaw 2026.6.8 just dropped.
💬 Richer Telegram WhatsApp
⚡️ Sharper agent & Gateway recovery
🧠 New models tougher memory
📊 Native /usage footers
📱 Smoother WebChat & iOS
Let’s dig in!
github.com/openclaw/openclaw…
25
39
311
19,345
Frank Y retweeted
13h
We're excited to join forces with @SpaceX to advance the frontier of useful AI. Expect significant improvements to Cursor soon.
SpaceX has exercised the option to acquire @cursor_ai in an all-stock transaction with the goal of building the world’s most useful AI models.
For the past few months, SpaceXAI has been jointly training a model with Cursor, which will be released in Cursor and Grok Build soon.
We look forward to working closely with the Cursor team to advance our frontier AI capabilities
1,014
2,619
26,003
2,642,608
should i keep awake???
14h
Fable 5 is temporarily unavailable
Anthropic changed the text from CURRENTLY to TEMPORARILY
are you feeling what am feeling here?
42
Frank Y retweeted
Jun 14
我的偶像郭璨 27 岁，SpaceX 任务控制室核心工程师，火箭发动机参数实时监测她说了算，异常时直接下中止指令
最离谱的是，她 20 多岁写的代码还在优化发射控制逻辑，把流程自动化提了一大截
结果一查工资，年薪现金才 13 万美元，硅谷白菜价
但人家持股 8–15 万股，按 135 刀 IPO 价算，身价直接干到 1500 万刀
之前纪录片里那个“花臂 吊带”出圈的，就是她
难怪网友喊她星舰女王，这履历放硅谷随便哪家都是抢着要，她偏选 SpaceX 硬卷
偶像，实力真是藏不住的
367
1,608
12,917
1,015,798
Frank Y retweeted
Jun 13
Check out Omnigent, an open source harness that lets you use all the existing code harnesses (Claude Code, Codex, OpenCode, pi), collaborate and share sessions in many modalities (e.g. Slack/Teams, cli, webui), while having a fine grained security model that really tightens the control on what agents can do/not do.
github.com/omnigent-ai/omnig…
9
53
315
20,615
So this is what I did when I see the tweet about Fable 5 shutdown. I asked Fable with max reasoning to draft design for future work as detail as possible , which allow lower intelligence agent can continue the work. Name the file called fable_legacy.md. Thank you, Fable 5. @AnthropicAI
3
97
what !??????
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
1
43
I spent the past week building a marketing automation harness as an OpenClaw plugin. It runs in production now, publishes real posts to a real client page, with an LLM in the loop.
My only rule from day 1: the model never touches anything irreversible.
some field notes 🧵
2
1
8
129
One footgun for OpenClaw plugin authors: a non-empty `plugins.allow` is a
strict allowlist, and it gates BUNDLED plugins too.
It silently took down my channel ingress for ~1 hour. Config looks fine, tool
catalog looks fine, but nobody home.
My daily runbook fails on this now. The trust anchor is
`plugins.entries.<id>.enabled`, not the allowlist.
1
1
41
This is not a framework. One plugin, one sqlite file, one level-triggered
loop, and a short list of invariants that a runbook re-proves every day.
Boring on purpose. Boring is why I can sleep while it publishes.
The model does the language. The harness does the consequences.
(next: pointing this at public traffic — a deterministic layer that kills
90% of inbound with zero model calls. that post comes when it's live, not
before)
34
Every publish is sanctioned in exactly one of two ways:
- a routed human approval, or
- a scheduled time gate, set at submit
never both.
And right before the POST, a verifier checks the whole lineage again: who
submitted, who owns it, caption verbatim, image sha256 pinned at submit,
anchor binding. A forged row refuses with zero provider calls.
1
20
The part that ate the most review rounds: exactly-once outbound.
- attempt marker written in its own transaction BEFORE the provider call.
never across HTTP
- readback first. same content without our marker = never claim it
- marker without record = stuck. surface it, never auto-retry
- uncertain = do nothing log
You can't unsend a public post. So when not sure, refuse. Retry is the
dangerous direction here, not the safe one.
3
24
OpenClaw 2026.6.6 🦞
🔒 Tighter security boundaries
💬 Safer Telegram iMessage delivery
🧠 Claude Fable 5 OpenRouter OAuth
⚡ Faster Control UI first replies
Less weird, more work done.
github.com/openclaw/openclaw…
230
85
797
67,719
Wow !
Jun 12
🌘 Kimi-K2.7-Code, our latest coding model, is now released and open-sourced!
🔷 Improved coding & agent performance over K2.6: 21.8% on Kimi Code Bench v2, 11.0% on Program Bench, and 31.5% on MLS Bench Lite.
🔷 Reasoning efficiency: Less overthinking, with 30% lower reasoning-token usage compared to K2.6.
🔷 Long-horizon coding: Improved instruction following, higher end-to-end coding task success rates.
⚡️ 6x High-Speed Mode coming soon!
🔌 Available today via Kimi API and Kimi Code.
🔗 Kimi Code: kimi.com/code
🔗 API: platform.moonshot.ai
1
1
82