Aetheris

Every PR at our agency gets reviewed by 5 AI agents before a human reads it. We just open-sourced the playbook 🧵

@BookitHQ Why it mattered: with no MLB games Jul 13–15, daily grading would have counted break days as missed picks and reset every active streak. Stray picks during the break can't void a streak either.

BookIt Sports 2026.7.11 has been released. Highlights: • Streaks are paused for the MLB All-Star break (Jul 13–16) - no picks needed, no streak resets •Picks reopen Friday, Jul 17 when the full slate returns • History and Leaderboard stay live through the break @BookitHQ

2 of the 3 run in any repo today. The third is wired to our internal tracker — it’s there as a reference implementation. Fork the pattern. MIT licensed: github.com/Aetheris-Solut…

Every PR at our agency gets reviewed by 5 AI agents before a human reads it. We just open-sourced the playbook 🧵

/fable-orchestrator — frontier models are expensive. So: the expensive model orchestrates and verifies, Opus builds, Sonnet implements to spec, Codex reviews every nontrivial diff. File-ownership matrices so parallel agents never collide.

/second-opinion — our favorite: Claude asks OpenAI’s Codex to review its work. Different model, different blind spots. Read-only sandbox, advisory only, never edits.

Every finding gets scored 0–100. ≥90 → posted as a blocker 70–89 → posted as before-merge <70 → discarded, never posted Unscored AI review is noise. The threshold is the whole game.

/aetheris-review — multi-agent PR review that knows what the ticket asked for. It pulls acceptance criteria from our tracker, then fans out 5 reviewers: bug hunt, git regression, spec compliance, prior review feedback, comment accuracy.

Context: we’re a software agency. Multiple client repos, small team, AI-heavy delivery. These are the Claude Code skills we actually run daily — now a public marketplace.

@haha_girrrl CS students who only learn to code are cooked. CS students who learn systems, distribution, and product are about to have the best decade of any generation.

Computer Science Students!! What's your backup plan if Artificial Intelligence writes better code than you in 2 years?

@Prathkum Code is the deliverable. Engineering is the judgment that decides what the deliverable should be. 2 completely different skills, finally being priced differently (& correctly)

AI created the illusion that writing code is all it takes to be an engineer.

@vineerpasam Zero users in a month isn’t a product problem, it’s a distribution problem. Too many ppl ship to nobody. Where were they supposed to find it?

You built a product, got 0 users after a month. Now what?

@Its_Nova1012 Own the customer, the data, or the distribution. Code was always the cheapest part of the stack.

Software Engineers, what’s your backup plan if Artificial Intelligence writes better code than you in 3 years?

Org charts in 2020: 200 boxes, 50 reporting lines. Org charts in 2026: 5 boxes, 50 dashed lines to agents. Org charts in 2030: 1 box. Management as a profession is about to get audited by reality.

Lots of people bought mac minis and gave it a $500 allowance to send calendar invites and provide an inbox summary

The next decade isn’t ‘AI replaces humans.’ It’s ‘humans get blamed for what AI did.’ Every org chart still ends in a person who signs the document, takes the call, and goes to court. That person just got 100x more leverage and 100x more risk. Pick your seat carefully.

Five years ago, leverage meant capital. Today, leverage means tokens. The cheapest unit of work in human history is a sentence typed into a model. We haven’t priced that into anything yet.

In 2020 the flex was ‘I work at Google.’ In 2023 it was ‘I’m a founder.’ In 2026 it’s ‘I have no employees and 8 agents.’ The status ladder is being rebuilt in real time and most people haven’t noticed.

Can't believe some people are still paying exuberant enterprise rates when we can build them a tailored solution in a month for pennies on the dollar