Play a game in your terminal while Claude thinks. Auto-pauses when Claude replies, resumes on your next prompt. Five titles: Break-In, Skyrun, Jetwash, Astros and Jaywalk.
★not rated 2 3mo agoA
tokens not measured
originalMIT
Real-time HUD statusline for Claude Code with API balance monitoring. Shows context health, tool activity, agent tracking, todo progress, git status, and DeepSeek/OpenAI/Anthropic balance tracking with low-balance warnings.
★not rated 2 2mo agoA
tokens not measured
originalMIT
A Korean-language safety loop for non-developers who ask an AI to keep repairing a task until it works. Its description says it currently supports a single repair mode with 14 safety guardrails enabled by default.
Describe what you want. Read one page. Approve four times. Get a product that does what the page says, proven live. One command: /compass. Four stages: Idea, Plan, Build, Go live. Fresh reviewers test every promise inside every stage.
★not rated 2
changed 10d agoA
tokens not measured
originalMIT
A cheap background model that watches your Claude Code session over its shoulder and, rarely, drops one short advisory into it — asynchronously, without ever blocking. It reads only the agent's emissions (29% of the transcript), never its tool results, so it is affordable to leave running for the whole session.
★not rated 2 2mo agoA
tokens not measured
originalApache-2.0
Reconnect MCP servers in Claude Code tmux sessions via automated key injection. No API required — works by driving the /mcp interactive menu through tmux send-keys.
★not rated 2 3mo agoA
tokens not measured
originalMIT
Implementation of the Ralph Wiggum technique - continuous self-referential AI loops for interactive iterative development. Run Claude in a while-true loop with the same prompt until task completion.
Swarm-orchestrated code refactoring and feature development with specialist agents. Supports multi-instance parallel agent spawning, blackboard context sharing, and interactive development gates.
★not rated 2 5mo agoA
tokens not measured
originalMIT
Statusline segment + prompt-submit guard for Claude Code's 5-hour and 7-day usage limits. Shows live usage %, and at a configurable threshold it blocks new prompts (with an override) so a long token-heavy task doesn't get cut off mid-run.
★not rated 2 4mo agoA
tokens not measured
originalMIT
Mines real online discussion (Reddit, Hacker News) for the verbatim language people use about a problem, so claims about customer pain answer real complaints instead of invented ones.
★not rated 2
changed 2mo agoA
tokens not measured
originalMIT