Seth Gammon @SethGammon

@DanielSmidstrup Understanding + Awareness + Articulation

@ZeBoris_ @gilgNYC This is in the chat section of the app, not the code section (I assume) as I'm on android as well

Yeah I was using the desktop app which is a drop down. As far as I'm aware if you start or customize the session on your computer and then use it on your phone app (which still connects to your computer) that works to use your set effort level. On the cloud container and on mobile it does not recognize the /effort skill ('/effort max' for example.) You might be able to have sub agents or spun up agents by Claude use higher effort levels by just asking it too. I haven't tested that though. Seems mobile cloud sessions are limited in that customization.

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.

@MASAKREofficial @Diablo Now that's:

@craigweiss A full circle moment

@_Rohan__mehta I've been making a lot of updates in this direction! Many recent, and more to come 🙏

Some of the improvements I've been working on for Citadel, the Operating System for Agents. 1. Local Citadel Control Plane Added a local operator dashboard path so Citadel state is visible without spelunking files. 2. Fleet DAG and Merge Steward Added dependency-aware Fleet task parsing and merge readiness checks. 3. Worktree Environment Readiness Profiles Added readiness reporting so isolated worktrees can be checked before high-autonomy work. 4. Signed Telemetry and Artifact Lineage Added hash-backed telemetry/artifact lineage and verification. 5. Memory Compiler and Semantic Blocks Added source-backed memory blocks for reusable project knowledge. 6. Phase Exit Evidence Contracts Added campaign evidence tables and a validator so phases cannot be marked complete from prose alone. 7. Sandbox Provider Contract Added an explicit sandbox-provider abstraction with supported worktree behavior and clear unsupported placeholders. 8. Skill and Workflow Packaging Upgrade Added skill metadata, cataloging, scaffold support, and packaging tests. 9. Thin Runtime Adapter Matrix Added runtime capability levels and tradeoffs so Citadel clearly says what Claude Code, Codex, OpenAI, or unknown runtimes can and cannot guarantee. 10. Shared Repo Map Substrate Upgraded `/map` into a reusable `core/map` substrate with file hashes, imports, exports, routes, package scripts, verification commands, scoped slices, and stale checks.

@rezoundous Bro I followed this rollercoaster with you. 100 to codex, then wanting to up it cause it was so good. Held off and got fable. Then upped that to max. Then pulled. 😂 Thank you for being there

@sattyyouneed Yes for certain. It wasn't out long so mostly you'll read anectdotes. Not all workflows and use cases are equal. The main takeaway I got was design was great, UI, UX, and articulating what I wanted felt easier to get good results. Gpt 5.5 is very good, but fable 5 was better.

@xiaoqiao6666666 Why do people keep setting up coins like this?

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.

Anthropic @AnthropicAI

23 hours ago

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

12K 25K 83K 78.9M 23K

Citadel: The Orchestration Harness for Agentic Development github.com/SethGammon/Cit…

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.

@zdogmode Harland David Sanders spent many years trying to sell his recipe, succeeding to open and franchise KFC in his 60s. If your friend doesn't want to be a founder that's completely fine. 35 ain't nothing. The only valid excuse to not try at something you want is death.

@dontflex00 They know about the product and have written posts/articles about it. Participated in a private preview competition last month. Overhauling ProofJudge (the project) for them to share as it runs on EigenCompute.

Sneak peek into something I'm demoing soon in connection with EigenCloud. Really loving the UI/UX, mainly this scan transition.

@alightinastorm @grok What a strong usage of Grok 😅👏

@0ii05i Thank you Captain

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.

Claude @claudeai

4 days ago

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.

5K 15K 105K 55.5M 22K