lightcap.eth 🏴

Heretic @ Codezero. (also CTO/CPO)

[lightcap.eth 🏴 @lightcap]

Credentials continue to be the target of just about every attack in 2026.

[SafeDep @safedepio]

🚨 The "𝙼𝚎𝚐𝚊𝚕𝚘𝚍𝚘𝚗" Campaign is live...

𝟻,𝟽𝟷𝟾 malicious commits to 𝟻,𝟻𝟼𝟷 GitHub repositories in a six-hour window.

Using throwaway accounts and forged author identities (build-bot, auto-ci, ci-bot, pipeline-bot), the attacker injected 𝙶𝚒𝚝𝙷𝚞𝚋 𝙰𝚌𝚝𝚒𝚘𝚗𝚜 workflows containing 𝚋𝚊𝚜𝚎𝟼𝟺-𝚎𝚗𝚌𝚘𝚍𝚎𝚍 bash payloads that exfiltrate:

- CI secrets,
- cloud credentials
- SSH keys
- OIDC tokens
- source code secrets

Check your repo / Technical details: https://t.co/tua3jYU1wa

[lightcap.eth 🏴 @lightcap]

@ScriptedAlchemy That's an interesting technique. Did you write the hook? I'd love to see it.

[lightcap.eth 🏴 @lightcap]

Did the security industry push for monocultures within orgs only to expose them to the potential of one threat that wipes them out? Same patch, same OSes, same apps. NSA/Mythos agricultural analog?

Bananas 🍌

[lightcap.eth 🏴 @lightcap]

@AustinBarack Models will commoditize at some point, it's a race to the bottom, leaving privacy as the primary moat for inference.

[fejau @fejau_inc]

Crypto industry seeing AI getting rugged by the government

[lightcap.eth 🏴 @lightcap]

Custom status lines also seem to be a potential culprit.

[lightcap.eth 🏴 @lightcap]

For anyone else that's having rendering issues with garbled text using Claude code with ghostty or tmux a mildly tolerable workaround is hitting ctrl+l (lower L) for a rerender.

[lightcap.eth 🏴 @lightcap]

Giving up on claude code TUI for a bit. The rendering bugs make it literally impossible to get meaningful work done.

Resizing the terminal window helps for the next turn's prompt but then goes right back to corrupted output.

[lightcap.eth 🏴 @lightcap]

@PaulPallaghy @mxcl This.

[lightcap.eth 🏴 @lightcap]

@mxcl The burnout is real. Take care of yourself and your family. You’ve given so much already. 🙏

[lightcap.eth 🏴 @lightcap]

@ClaudeCodeLog Dear god please let it unfuck my tui rendering.

[lightcap.eth 🏴 @lightcap]

One thing I've noticed today is that Fable is *much* better with skills loading and usage. I rarely have to remind it to use an intended skill, it's been picking them all up just as part of the conversations.

Opus needed constant reminders. This is a huge improvement to me.

[lightcap.eth 🏴 @lightcap]

Maybe the @ClaudeDevs team can use Fable 5 to fix the constant rendering issues? Please?

[lightcap.eth 🏴 @lightcap]

Claude Code unusable in every session today.

[jpark @jparkjmc]

new policy from anthropic: if you use fable/mythos, they collect your data.

no exceptions. not even for enterprise partners.

[lightcap.eth 🏴 @lightcap]

Anyone having more rendering issues than normal with claude code tui? /cc @ClaudeDevs

[lightcap.eth 🏴 @lightcap]

What's going on with @Bankless support? I've been unable to cancel my account and it's been 20 days with no response. Not a good look.

[lightcap.eth 🏴 @lightcap]

Fable 5 will require usage credits in two weeks. Slowly and then all at once moving us all to usage from subs.

Real opportunity for OpenAI here if they can keep providing the subs with their next model upgrade.

[lightcap.eth 🏴 @lightcap]

Careful out there kids. This phishing ad keeps coming back. Keep those seed phrases safe.

[lightcap.eth 🏴 @lightcap]

Dozens of references to CISOs helplessness to stop the org-wide rollout of Claude.

[lightcap.eth 🏴 @lightcap]

One single mention of OpenAI over a day's programming at the enterprise focused AI cyber event I just attended.

Claude mentioned in just about every other sentence.

Anthropic owns the enterprise.