Anthony Adams

Something that surprised me while building this week: "OpenAI-compatible" is a statement about the API shape, not about how code edits are expressed inside it.

Aider does not use tool calls. It sends edits as structured text blocks SEARCH/REPLACE delimiters inside the assistant message. The filename is on the line before the fence, not inside it. Cline and Continue use proper tool calls. A response from some backends carries both.

The adapter we shipped this week (`openai_chat.py` in the LineageLens proxy) handles all three patterns: tool-call edits, text-content edits (Aider SEARCH/REPLACE, unified diffs, fenced blocks), and mixed responses where both appear together. It covers Aider, Cline, Continue, Copilot CLI, Goose, Windsurf, and any OpenAI-compatible backend Azure, groq, fireworks, mistral, together.ai.

The hardest design decision was not the parsing. It was fail-open vs. fail-safe. If capture logic throws an error, do you surface it (fail-safe) or swallow it and pass through (fail-open)? We chose fail-open. A proxy that occasionally interrupts a developer's coding session gets uninstalled. Governance with a 60% retention rate is strictly worse than governance with 95% retention.

Hi everyone!

News #1: ArkTabs crossed 2000+ users with 96% userbase in US.

News #2: Arkida Mockups is launching on Product Hunt this Monday! https://www.producthunt.com/prod...

Follows, comments and shares would be highly appreciated! The more popularity it get, the more features it will get.

OpenAI just made AI memory automatic.
The problem? you can no longer fully audit what it remembers about you.

Dreaming V3 (launched June 4) doesn't ask you to save memories anymore. It runs in the background, reads across years of your conversations, and synthesizes a profile of you, and automatically updating as your life changes.
It remembered your Singapore trip was in the future. Now it knows you went. You didn't tell it that. It figured it out.

The audit trail problem: TechTimes reported the update "limits" what users can actually see. You used to have a list of saved memories you could read and delete. Now you have a synthesized profile that's broader than that list, and you can only partially inspect it.
A CHI 2026 study called it the "personalization-convenience paradox": the feature users value most is also the one they can least control.