Always-on working discipline for every project: report status and ETA for long-running tasks when they finish, fail, or stall, and tag each project with its categories so domain plugins such as research load only where they belong, put every web service a machine hosts behind one token-gated entry point, and keep a shareable per-user GPU niceness counter for the shared behemoth box.
Plugin core from marketplace dansolombrino-skills in dansolombrino/skills. Commands read from the repo's manifest — today.
behemoth-gpu-nicenessRun and share the niceness counter for the shared behemoth GPU box - a sampler that records, every minute, which user holds which GPU, and a read-only page behind service-hub that ranks users by the GPU-hours they used, with each one's share of all GPU time, over 24 h, 7 d, 30 d or everything kept.
progress-updatesMake a long-running task measurable and report it when something changes - file transfers (rsync, scp, cloud copies), downloads, single training or evaluation runs, builds, or any background command expected to take more than a few minutes. Use when such a task is started or is already running and b
project-initTag a new or existing project with its categories in .project.toml and generate each host's per-project plugin enablement from it, so domain plugins such as research load only in the projects that use them. Use when the user asks to set up, initialize, tag, or categorize a project, to make a repo a
service-hubRun one web entry point per machine that lists every hosted web service on a landing page and serves each at http://host:port/service/ through a token-gated reverse proxy, so the backends listen on loopback only. Use when the user wants to put dashboards, viewers or other web apps behind one port or
token-usageRun the token-usage page - an indexer that reads the session logs the coding agents (Claude Code, omp, Codex, including the ones Orca launches) already write on this machine and, over ssh, on the rest of the fleet, and a read-only page behind service-hub that charts every model call like a stock - t
Materialize reproducible experiments across verified GPU rigs: scaffold projects, provision exact environments, deploy tested revisions, dispatch GPU waves as a self-balancing pull queue, supervise them from the hub without an open chat, keep a fleet-wide GPU lane board, and visualize results.
12 skills · 2 agents · hooks
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
15 skills · hooks
Scaffold, validate, and publish new Claude Code plugins with proper structure
3 skills · 1 command · 1 agent
Up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your coding agent's context.
1 MCP server
Use Alibaba Open Code Review from local Codex.
2 skills
Claude Code skill for creating and presenting Slidev slidedecks
1 skill
Code intelligence powered by a knowledge graph. Provides execution flow tracing, blast radius analysis, and augmented search across your codebase.
12 skills · 1 MCP server · hooks
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.