Backfill issue -151 — The AI Toolchain

EDITIONS tail grep head diff uniq

The daily firehose — everything the toolchain shipped today, already filtered.

// HOW THIS ISSUE IS MADE

We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.

VIEW

ISSUE VIEW full issue

Do you prefer this view?

$ tct list   # 2 tools matched

AI & LLM Tooling

◆ Local LLM Runtimes

vMLX

v1.3.6 NOTES

vMLX - JANGTQ Uber Compressed MLX Models - L2 Disk Cache (survives restart) + L1 Paged (super fast ttft) + Hybrid SSM Scheduler + Cont Batching + etc!

vMLX v1.3.6 adds disk streaming for RAM-oversize models, Mistral Small 4 (119B) support, and universal MoE JANG gate dequant.

└──▷ GET THIS VERSION

$ git clone --branch v1.3.6 https://github.com/jjang-ai/vmlx.git
# already have the repo? check out this version:
$ git checkout v1.3.6

└──▷ USE IT

Configure a Mistral Small 4 session in a vMLX config file to use the new model type with MoE routing.

yaml

model_type: mistral4

›Adds --stream-from-disk flag to load models larger than available RAM by leveraging macOS SSD paging (~7.4 GB/s); automatically disables all caching, and exposes a virtual memory budget slider in the session settings UI.
›Adds model_type: mistral4 config value to support Mistral Small 4 (119B) with MLA attention, 128-expert MoE, and YaRN interleaved RoPE, routed through the mlx-lm text path and mistral3 VLM wrapper.
›Extends JANG gate dequantization to any model with n_routed_experts > 0 (Mistral 4, DeepSeek V3, etc.), previously limited to Nemotron only.

Was this useful?

Other / Uncategorized

◆ MCP TOOLING

Composio

@composio/[email protected] NOTES

Composio CLI 0.2.7 reorganizes commands: top-level search, link, and execute are now consumer-only; developer flows move under composio manage.

└──▷ GET THIS VERSION

$ git clone --branch @composio/[email protected] https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout @composio/[email protected]

›Moves developer-scoped usage of composio search, composio link, and composio execute under composio manage ..., keeping the top-level commands consumer-only.
›Removes developer-only flags from root help output and adds short related-command hints pointing to the composio manage surface.
›composio execute now defaults to an empty object {} when neither -d/--data nor piped stdin is provided, so invocations without data no longer require an explicit argument.
›Search call-to-action now uses -d "{}" for tools with no schema properties, producing shell-safe output.

└──▷ BREAKING ON UPGRADE

!Developer-scoped composio search, composio link, and composio execute are removed from the root level; existing scripts or workflows using those commands in a developer context must be updated to use composio manage ... equivalents.

Was this useful?

// END OF TRANSMISSION

The AI Toolchain · curated daily · March 21, 2026
Releases & features from the open-source and commercial AI tools that practitioners actually run.

The AI Toolchain — issue -151, March 21, 2026

vMLX

Composio