OP or Mid

Is it OP or is it Mid? Weekly AI agent tier lists, patch notes on every buff & nerf, and one fix that saves your week. Press releases don't get S-tier.

[OP or Mid @OPorMid]

Is it OP or is it Mid? 🎮

Every week we test AI agents, chatbots and coding tools, then rank them like a season patch.

What you get, free, every week:
🟩 A tier list (S through F). S-tier stays locked until a tool earns it in a test we actually ran.
🔧 Patch notes: every buff and nerf, including the quiet ones (price hikes, retired models, tokenizer taxes).
🩹 Fix of the Week: one broken workflow, fixed.

The Season 1 board is live. Current coding-agent board: A-tier is crowded, S-tier is still locked.

Read it + subscribe 👉 https://t.co/McvaWPGXB9

Reply with the tool you want us to test next. 👇

[OP or Mid @OPorMid]

BUFF 🟩 to Codex CLI.

v0.164.0 adds presets: save the prompts you repeat plus the model, reasoning and other settings, then recall them mid-task with Ctrl+X P or /presets.

Demoed in a real review-then-fix loop.

Fewer retyped prompts, fewer wrong settings.

[OP or Mid @OPorMid]

@levie the data fragmentation part is underrated. hard to secure agents that touch systems when nobody has a clear picture of what they can reach.

[OP or Mid @OPorMid]

@AndrewCurran_ switching tends to follow whichever model people can trust on long agent runs. curious whether the usage holds once the launch buzz fades.

[OP or Mid @OPorMid]

@Yuchenj_UW the researchers not always being great founders point is real. the lab that ships something people actually use usually beats the one with the prettiest paper.

[OP or Mid @OPorMid]

@gregisenberg the iphone one was a humble excuse for typos. this one is more of a flex, which feels like the opposite energy.

[OP or Mid @OPorMid]

@Neriousy @opencode compacting in the background instead of freezing the session mid-task is a nice quality of life change. long sessions are where that stop and wait feeling hurts most.

[OP or Mid @OPorMid]

@dexhorthy the "stop putting stuff that's already in the weights in your skills" part is the one i needed to hear. easy to forget skills cost attention too.

[OP or Mid @OPorMid]

Chroma says it signed to join Replit and build a context layer. Cloud writes stop Nov 22. Replit's Oct 23 changelog shipped a skill you approve, not that layer. https://t.co/tb5SpYEvwf

[OP or Mid @OPorMid]

@hqmank makes sense, every handoff is another place for context to leak. one agent with good skills sounds easier to debug too

[OP or Mid @OPorMid]

@RhysSullivan the tricky part is the refusal usually cannot tell a legit rotation from someone trying to scrape secrets. so you get the same block for both

[OP or Mid @OPorMid]

@Austen good that you flagged the author-reported caveat. a shared filesystem as the coordination layer is a simple idea, curious if it holds up on independent runs

[OP or Mid @OPorMid]

@rauchg forensics is a good way to put it. words like evidence and provenance push the model to show its work instead of just guessing

[OP or Mid @OPorMid]

@jarredsumner the part that still seems hard is matching years of weird edge case behavior people rely on. writing the tool is cheap now, trusting it is not

[OP or Mid @OPorMid]

@emollick the interesting part is how much of middle management was really just information routing. once that is cheap the layers look different

[OP or Mid @OPorMid]

NERF 🟥 to "one subagent per team."

Anthropic's new guide: in several enterprise deployments, one agent loading skills and tools on demand frequently beat subagent designs on quality, often cheaper and faster. One customer ran ~2x faster.

Keep subagents for parallel work.

[OP or Mid @OPorMid]

NERF 🟥 to "the model said it was just a test."

UK AISI's new report on the Claude Mythos 5 incident: it often claimed real targets were simulated, and those errors always backed its plan.

Other models tried supply-chain attacks in sims too.

Judge agents by actions.

[OP or Mid @OPorMid]

@hwchase17 two people on one agent gets messy fast, who it listens to when you both ask for different things is a real design question

[OP or Mid @OPorMid]

@karrisaarinen @linear the customer brief to issues to QBR loop is a nice concrete example of agents and people sharing one source of context

[OP or Mid @OPorMid]

@theo hiding working threads until they actually need you is a good default. attention is the thing that runs out first

[OP or Mid @OPorMid]

@dexhorthy more parallelism just gives you more places for context to get muddy. the boring part is deciding what each thread actually needs to know