Joined September 2011
- Tweets 2,560
- Following 588
- Followers 556
- Likes 14,632
135 Photos and videos
Pinned Tweet
Jun 9
Citadel just crossed 600 GitHub stars.
Citadel is my agent orchestration harness for Claude Code Codex with repo-native memory, session-persistent campaigns, and parallel agents in isolated worktrees.
If you want the smoothest experience using Codex and Claude Code together, check out the repo below.
Really appreciate everyone who starred it, tried it, broke it, and helped sharpen it.
16
1
30
2,503
hey big head @DarioAmodei
when's fable coming back? :)
41
12
569
73,952
Awaiting impatiently
i’m hearing positive talks about fable being given back to the people (globally)
expecting news to break imminently.
48
Jun 13
Dev friend of mine asked if having this error screen was practical for my project some months back.
I cannot count the number of times it allowed me to instantly zero in and fix the problem.
Build custom tools to unlock clarity on your work.
97
Jun 13
A loud voice in the discussion is saying it's Anthropics fault for marketing the dangers of this latest model.
What we know is that someone reported a jailbreak and their concerns.
Maybe the report came influenced by their marketing of day zero exploits.
Maybe backroom deals and sneaky competition is to blame.
The government is notoriously slow to understand technology, and are likely overreaching.
Fable 5 was a powerhouse, and I hope it returns soon.
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…
5
4
747
Jun 11
I love when people lend their suggestions, always amazing when someone is using what you made and enjoying it enough to contribute their ideas.
Check out Citadel below, great with Claude Codes new Fable 5, and (hopefully) GPT 5.6 when it comes out! Already great with Codex.
1
7
1,364
Jun 11
Citadel: The Orchestration Harness for Agentic Development
github.com/SethGammon/Citade…
2
3
338
Jun 10
Sneak peek into something I'm demoing soon in connection with EigenCloud. Really loving the UI/UX, mainly this scan transition.
2
10
446
So I was totally sold on GPT 5.5. Solely on the basis of what I saw tonight with Fable, I will start to use again Claude extensively, probably as the main model as long as Fable is accessible. It's not just very capable, it performs organized steps and have deep understanding.
59
29
1,139
80,633
Jun 9
I am a little shocked that Fable 5 didn't one shot my usage limits.
The quality is definitely 'better' but I'm unsure if that's narrow or across the board. New model smell will always influence toward impressive.
Looking to get into it more.
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
Its capabilities exceed those of any model we’ve ever made generally available.
3
8
386
Jun 9
Excited to see how it feels coding
Today, we're introducing Claude Fable 5 and Mythos 5, two configurations of our next major language model.
I'd normally highlight the numbers: It's SOTA on nearly all benchmarks. I want to talk about something else, because with Fable 5 out in the world, I think a third era quietly started today.
I lead Claude Code & Cowork on the desktop, so I think a lot about how people use AI to get work done. I believe we're about to see a major shift, moving from giving AI tasks to giving it responsibilities.
1
177
Jun 9
Direct prompting is asking an agent to do a task.
A loop is the structure around repeated work: what triggers it, what input it uses, what it is allowed to change, how it verifies progress, when it retries, when it stops, and what evidence it leaves behind.
Citadel already had looping workflows. This pass made them more formal and inspectable: loop contracts, status surfaces, bounded /loop runs, templates, shared stop states, verifier budgets, and evidence trails.
The goal is not “prompt the agent harder.”
It’s to make the operating loop clear enough that agents can work inside it without the user manually steering every turn.
Shared below is an article I wrote about going from zero to fleets of agents (which covers a bit on workflows that would contribute to loops) as well as the Citadel OSS repo.
Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore.
You should be designing loops that prompt your agents.
2
5
556
Jun 9
3
83
Jun 9
Citadel just crossed 600 GitHub stars.
Citadel is my agent orchestration harness for Claude Code Codex with repo-native memory, session-persistent campaigns, and parallel agents in isolated worktrees.
If you want the smoothest experience using Codex and Claude Code together, check out the repo below.
Really appreciate everyone who starred it, tried it, broke it, and helped sharpen it.
16
1
30
2,503
Jun 9
2
2
209
Seth Gammon retweeted
Jun 8
if you’re still writing loops that prompt coding agents you’re falling behind. you need to build a meta agent that infers what loops you would have wanted based on your vibe and then write those loops
261
233
4,593
202,402
Jun 5
4
1
9
536
Jun 4
Enhanced Token Info for Citadel will be available on @dexscreener shortly!
Thank you for those that are interested in this sort of vehicle to drive attention and funding to the project.
56
2
88
3,814