AI tools
A full dev methodology as auto-triggering skills (works across agents).
Anthropic's agentic coding CLI — delegate coding tasks from the terminal.
Installable bundles of skills, MCP servers and commands for Claude Code.
tmux for coding agents — real terminal panes, blocked/working/done state at a glance, persistence on detach, SSH from your phone, and an API agents can drive.
Proprietary desktop IDE (VS Code-based) that runs AI agents in parallel, each in its own isolated Git worktree, with cross-agent orchestration and Jira/GitHub review.
A minimal, aggressively-hackable coding agent for the terminal.
Loop the same prompt into a coding agent until the task is done — brutally simple autonomous agentic development.
AI-first code editor with an agent built into the IDE.
Lightweight spec layer — agree on what to build before code, brownfield-friendly.
Governed spec-driven workflow + multi-agent worktrees for a software factory.
A 70-line CLAUDE.md of anti-overengineering discipline.
Small, composable agent skills for everyday coding.
218 specialized subagent personas across 15 divisions, installable into most agent tools.
A one-file 'lazy senior dev' skill — makes agents pick the simplest solution that works.
Senior-engineer workflows as portable agent skills.
Portable skill that rewrites text to remove the 33 tells of AI writing (em dashes, rule-of-three, 'let's dive in') — based on Wikipedia's Signs of AI writing.
Package manager + registry for agent skills; find-skills lets your agent discover and install them.
Persistent memory for Claude Code across sessions.
Gives a coding agent a design brain for UI/UX work.
A dozen modular anti-slop frontend skills with numeric dials for variance, motion and density.
A design language for agents: 23 commands, a 58-check slop detector, and hooks that can fail a PR.
One opinionated design skill — builds, audits, redesigns and studies, and refuses to repeat its own structures.
Structured defensive-security playbooks for AI agents.
Curated directory of Claude Code skills, hooks and commands.
Makes coding agents stop padding — fewer output tokens, same answers.
Local-first compression layer — shrinks tool outputs, logs, RAG and history before they hit the model; reversible.
Local TUI dashboard — see where your AI coding tokens and cost go.
Lean CLI for Claude Code token usage and estimated cost.
The most-adopted framework for building LLM apps and agents.
One command to give any agent web access — picks, installs and routes the best free backend per platform.
One OpenAI-format API (and self-hosted gateway) for 100+ LLMs.
Multi-agent (tree-sitter + LLM) plugin that turns a codebase into an interactive knowledge graph you can search, tour, and ask questions about.
Local hybrid search over markdown files (CLI + MCP), on-device.
Local-first, plain-Markdown knowledge base — free, extensible, and yours; the power-user's second brain (and an ideal LLM knowledge source).
Turns a codebase (code, docs, papers, images) into a queryable knowledge graph.
Turns Claude Code into a job-search pipeline — scores offers, tailors CVs, scans portals.
Run open LLMs locally with a single command.