An open large language model pairing DeepSeek Sparse Attention (DSA) for cheaper long-context inference with a scaled RL pipeline. Authors claim parity with GPT-5, with a high-compute Speciale variant surpassing it and rivaling Gemini-3.0-Pro on reasoning.
AI Agent2025
A configuration layer for the OpenCode and Codex CLI coding agents. One install adds specialized sub-agents, lifecycle hooks, and bundled MCP servers — web search, docs lookup, code search — turning a bare agent into a harness for large codebases.
AI Agent2025
Orchestrates multi-model LLM agents and developer workflows as an OpenCode plugin — runs background specialists, LSP/AST-aware refactors, hash-anchored edits, and built-in MCPs. Designed for agent-driven code automation and multi-model orchestration.
AI Agent2025
Coordinates about a dozen role-based AI agents — analyst, architect, developer, QA, scrum master — through a CLI, taking a feature from PRD and architecture docs into an automated dev cycle. Runs inside Claude Code, Cursor, Codex, or Gemini.
AI Agent2025
Provides a CLI-first framework to orchestrate autonomous AI agents and development workflows. Includes role-based agents, the ADE execution pipeline, IDE hooks and an NPX installer for quick setup—best for teams automating planning→development→QA.
AI Agent2025
Provides a Plan→Work→Review→Release harness for Claude Code agent workflows that enforces spec-driven tasks, TDD-backed implementation, independent review, and packaged evidence for PRs/releases. Exposes plugin/CLI commands and a Go-native guardrail engine to keep agent-driven code delivery reproducible and auditable.
AI Agent2025
Coordinates multiple AI coding agents and persists work state in git-backed hooks; provides convoy-based work tracking, an AI coordinator (Mayor), agent lifecycle/watchdog tooling, and a merge/refinery workflow for reliable multi-agent code work.
AI Agent2025
Packages reusable agent capabilities as lightweight 'skills' (folders with a SKILL.md) that capture procedural knowledge and workflows; uses progressive disclosure so agents load minimal metadata at discovery and fetch full instructions and resources only when needed.
AI Coding2025
Equips AI coding assistants like Claude Code and Cursor with 75+ executable tools, an MCP server, reusable skills, and a Python library to build on Databricks—Spark pipelines, jobs, dashboards, Unity Catalog resources, and ML workflows—from your editor.
AI Infra2025
Runs untrusted AI-agent code, commands, and file operations inside isolated sandboxes that scale from local Docker to Kubernetes. One Sandbox Protocol unifies both runtimes, with gVisor, Kata, and Firecracker isolation and SDKs across five languages.
Fifteen reusable agent skills for curating LLM context windows, treating attention decay—not token capacity—as the real constraint. A routing layer benchmarked at 0.92 top-1 accuracy selects the right skill for each task.
AI Agent2025
Runs LLM-driven coding agents as durable workflows on Vercel, orchestrating isolated sandbox VMs for repo work, shell and file tools, and optional auto-commit/PR flows — designed for cloud-hosted, resumable developer automation.