M-Pineapple 🍍

Open-source MCP servers & Swift tools for Claude, Cursor and macOS. Building in public.

[Cursor @cursor_ai]

Cloud agents can now build interactive canvases.

Ask for a dashboard or report while the agent works, then share the finished canvas with your team as a link.

[ClaudeDevs @ClaudeDevs]

We keep improving performance across Claude, and you may have noticed it starts responding sooner!

Claude Code now gzips the conversation it sends on every turn, so it reaches the model faster. Cloud sessions also start up as soon as you start typing.

[M-Pineapple 🍍 @M_Pineapple__]

@nullbnd @nullbnd Yep. Heartbeat with a count. "0 fires across 12 runs" is useful. "Nothing on fire" still works when the watcher slept through the week.

[M-Pineapple 🍍 @M_Pineapple__]

Vercel paid out $242k to people who escaped its sandbox.

My coding agent escapes mine for free every time I click "always allow".

[M-Pineapple 🍍 @M_Pineapple__]

@nullbnd Quiet but not mute. One line a night in week one, something like "still learning your chaos, nothing on fire". Total silence and I would assume it crashed, which is its own weird thing.

[Anthropic @AnthropicAI]

New from our STEM Fellows Program: Designing more efficient quantum circuits with Claude.

Given plain-language instructions, Claude Mythos 5 designed quantum circuits that improve on the best published ones, sometimes by an order of magnitude or more.

More efficient circuits reduce the resources needed by quantum algorithms, which could bring quantum computing applications in chemistry and materials science within reach sooner.

Read more: https://t.co/QCvoJLDmBo

[OpenCode @opencode]

GLM 5.3 Flash is now free in OpenCode

[M-Pineapple 🍍 @M_Pineapple__]

Running local models is easy. Knowing what they're doing is harder.

msty-admin-mcp 6.1.5: 56 tools for Msty Studio in any MCP client. One pip install, plus a calm security harden.

https://t.co/PhuYaZ83G1

#MCP #LocalLLM #AI #OpenSource #Ollama

[M-Pineapple 🍍 @M_Pineapple__]

Opus 5.5 fast mode is out. Same model, 2.5x faster, twice the price.

Finally I can pay double to be told "you're absolutely right" in under a second.

[Dan Kornas @DanKornas]

Your AI agents shouldn’t lose the thread every time you switch tools.

Kenfold is a self-hosted memory server for builders using Claude Code, Codex, OpenCode, ChatGPT, and local models.

It helps you carry project context from one tool to the next by sharing typed memories over MCP, then loading relevant context at the start of supported agent sessions.

Key features:

• Typed memory – separates semantic, episodic, project, preference, codebase, and temporary memories with distinct lifecycles
• Cross-agent handoffs – leave a note in Claude Code and pick it up in Codex
• Session hooks – load repository context at session start and record local session summaries for supported agents
• Hybrid retrieval – combines full-text and vector search, with optional reranking when configured
• Controlled memory changes – proposed extracted memories require approval, and replacements retain prior history

It’s open-source (MIT license).

Link in the reply 👇

[ClaudeDevs @ClaudeDevs]

Opus 5.5 fast mode in the Claude Platform is no longer in research preview, and we’ve removed the waitlist!

It’s the same Opus 5.5 but up to 2.5x faster, at $8 / $40 per million tokens.

It's also available in Microsoft Foundry (Hosted on Azure), with Google Cloud coming soon.

Fast mode in Claude Code remains unchanged, and uses extra usage credits.

See our docs here:
https://t.co/RikPqqIY0W

https://t.co/oB0fhEgC88

[Mercury @mercury]

Your agents can now take action via Mercury MCP

[ClaudeDevs @ClaudeDevs]

Code Review is now generally available for Claude Team and Enterprise plans.

Reviews are now faster, catch more of the bugs that matter, and admins can pick from Quick, Standard, or Deep review levels for each repo.

[M-Pineapple 🍍 @M_Pineapple__]

@nullbnd Only if it fits on one screen. What ran, what touched anything that matters, and the one weird thing. Any longer and it joins the rest in the log graveyard.

[M-Pineapple 🍍 @M_Pineapple__]

@cataofficial85 "Cleanup" is the scariest word in any agent log. Every delete should have to say its real name before it runs.

[M-Pineapple 🍍 @M_Pineapple__]

@nullbnd Honestly yes. Scoped to one repo with a 24h expiry, "always allow" turns into "allow today", which is about how long I trust my own 2am judgment.

[TechSpot @TechSpot]

Anthropic ditched Git for its own version control system because Claude is writing too much code

Scion system cuts code checkout times from minutes to seconds compared to Git
https://t.co/flaS26bddH

[Vercel @vercel]

Run your backend on Vercel.

• Python is generally available
• Containers for other runtimes & system libraries
• Daemons for async jobs & background workers

https://t.co/spPCQWUaWc

[M-Pineapple 🍍 @M_Pineapple__]

@cataofficial85 Exactly. Let it read anything, make it ask before it deletes anything. rm -rf always gets a human.

[M-Pineapple 🍍 @M_Pineapple__]

Your HubSpot is messier than you think.
Ask your AI for an audit and get duplicates, stale deals and broken workflows flagged in seconds. Works with Claude, ChatGPT, Cursor and any MCP client.
https://t.co/qv2EzDHJaN
#MCP #AI #OpenSource #HubSpot

[M-Pineapple 🍍 @M_Pineapple__]

@cataofficial85 Agreed. The allowlist is the seatbelt. "Always allow" is me unbuckling at 2am because the agent asked nicely.