TL;DR: AdaL 1.8.4 adds write-like-human, a capability that rewrites anything meant for a person to read so it stops sounding like a model wrote it. BYOAK now takes a base URL alongside the key, so AdaL can run against your own gateway. And /usage shows your weekly usage and when the window resets. Since then, Claude Opus 5.5 and Grok 4.7 joined AdaL at 50% off, and GPT-6 Astra is supported for the ChatGPT subscription.
No one reads on AI slop
Your agent writes a lot of text that other people read. PR descriptions, commit messages, release notes, docs, outreach, posts. Most of it arrives with the same fingerprints: an em dash in every other sentence, three parallel examples where one would do, a closing line that repeats the paragraph above it, and the construction that gives the whole thing away, "it's not just X, it's Y".
Readers notice. A PR description that reads like a model wrote it gets skimmed instead of read, and the reviewer trusts the summary less than they would trust two plain sentences from you.
The capability
write-like-human is a capability: ask AdaL to write like a human, or hand it a draft to fix, and it loads itself mid-session.
What loads is an editing procedure, not a tool, and it writes in your voice. Give it a sample and it copies your sentence length and punctuation; without one, it matches the kind of text, so essays keep their opinions and social posts go casual. It only rewrites text meant for other people, not AdaL's replies to you, and it stays loaded through compaction, so a long session keeps it to the end.
Point AdaL at your own endpoint
BYOAK used to take a provider and a key. Now it takes a URL as well:
/byoak add <provider> <api-key> <base-url>
Requests for that provider go straight to your URL with your key. They do not pass through AdaL, and they take effect on the current model with no restart. The /byoak dialog shows the URL next to the key, so you can always see where your requests are going.
This is for teams that route AI requests through a company gateway, and for anyone running a model on their own machine. One catch: your endpoint has to accept the same model name AdaL asks for. Gateways and proxies that pass requests straight through do.
Full setup is in the BYOAK docs.
Three new models
Claude Opus 5.5 and Grok 4.7 are in AdaL at 50% off through September 29, and GPT-6 Astra is supported for the ChatGPT subscription. Pick any of them with /model.
- Claude Opus 5.5: Anthropic's newest Opus, with a 1M-token context window.
- Grok 4.7: xAI's latest, with a 500K-token context window and a new
xhighreasoning effort. - GPT-6 Astra: supported for the ChatGPT subscription, so it runs on your own ChatGPT plan instead of AdaL credits. Availability depends on your plan. See ChatGPT Subscription to connect.
Know where you stand
You can finally see how much of this week is left. /usage shows your plan, your weekly usage, when the window resets, and your credit balance, in the CLI and the desktop app.
You also get a warning at 70%, 80% and 90%, so you know well before your usage runs out.
Also in 1.8.4
- A clearer student discount. Students can claim the discount with our student program.
- Clickable pull requests. The PR number shows on the bottom right of the terminal. Click it to go to the PR.
- Steadier tool calls. Improved tool-call stability across several models, and ChatGPT Plus and Pro models can run tools again.
Try it on your next draft. Open AdaL (it updates itself), paste something you are about to send, and ask it to "write like a human". Then read the two versions side by side.
Full notes are in the changelog, and how capabilities load is in the docs.