Trending repositories

Ranked by Attention Score: stars gained over the last 7 complete days

Period

Sort order

Tag

Filters (1)

Apply

Filters

Primary language

License

Total stars

Last push

Created

Archived

Clear all filters

6 repositories by Attention Score · Oct 28, 2026 – Nov 3, 2026 (UTC)

Language: Rust
Tag: Agent sandbox
Clear all

1.trycua / cua

Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.

apple
cua
lume
macos
virtualization
virtualization-framework
+12 more

Rust MIT32K total2.3K forks

+565 stars

from 31,112 · score 6.34

Ephemeral Workspaces, durable shared history. LayerFS gives every agent a complete, writable filesystem Workspace over shared committed history, without copying the base. Agents work in parallel, commit what changed as deduplicated checkpoints, then branch, rewind or merge. Try it with LayerFS × DSH.

agent-infrastructure
ai-agents
checkpoint
content-addressable-storage
copy-on-write
deduplication
+10 more

Rust MIT2.7K total143 forks

+280 stars

from 2,379 · score 5.64

3.pgrundev / pgx

PGX — Lightweight Postgres for AI agents. Written in Rust. Thousands of isolated databases in one runtime🐘

agents
database
postgres
postgresql
rust
rust-lang
+4 more

Rust AGPL-3.0654 total33 forks

+156 stars

from 498 · score 5.06

4.delta-box / deltabox-lite

Lightweight, agent-native sandboxing with diff, commit and abort

agent-sandbox
dsh-plugin
harness
pi-plugin
sandbox

Rust APACHE-2.0306 total19 forks

+103 stars

from 203 · score 4.64

5.perplexityai / SPACE

Sandboxed Platform for Agentic Code Execution: pluggable-runtime sandboxes for running agent and user code in isolation.

Rust APACHE-2.02K total183 forks

+63 stars

from 1,982 · score 4.16

6.microsoft / nvx

Cross-Platform Micro-VM Sandbox for Agentic Workloads

agentic-coding
agentic-engineering
agentic-workflows
microvm
sandbox
hyperv
+4 more

Rust MIT813 total53 forks

+29 stars

from 784 · score 3.40

Order: Attention Score

Most net stars gained in the window first. The score only puts this gain on a log scale, so the order is the order of the gain.

How the ranking works

Ranked by stars gained in the window. The score is the natural log of one plus that gain, so a repository's rank is exactly its star gain's rank.

score = log1p(net stars gained in the trailing T complete UTC days)

A repository is ranked only when its Star History covers every day of the window and its latest Observation is not missing.

The longest window is 180 days. Star History from the Repository Host goes back at most 180 days before a repository's first Observation. A longer window needs more days of Observations first. A repository younger than the window is not ranked in it.

The filters narrow the same eligible repositories. A repository matches the topics when it has any of them. Total stars is the count now. Last push and creation on the Repository Host count back from the start of today (UTC). The counts beside the languages, licenses, and topics include all ranked repositories in the window. A Tag holds the repositories that have it; see Tags.

Repositories enter the index from the Seed Dataset or by gaining stars faster than the published admission threshold, whatever their size. Admission is permanent, so an archived repository can still rank.