Build engineering digital twins as code with AI agents: one source-of-truth design.json mirrored into the build, a cited material-property source of truth (materials.json — never hallucinated allowables), every requirement gets an executable scored audit (PASS/WARN/FAIL), uv PEP723 scripts, rerun visualization (toward.
★not rated 2 5mo agoA
tokens not measured
originalMIT
Idea-to-playable harness for Expo (React Native + TypeScript) apps and games: research, plan, design, contract, and a generator-evaluator build loop with skeptical QA. --kind app|game.
★not rated 2
changed 22d agoA
tokens not measured
originalMIT
A review layer your agent applies to its own HTML or Markdown output in one command. You pin feedback to the exact word or element, and Markup Mode compiles each note to Markdown precise enough for the agent to act on, then hands it back. No backend, no build step.
★not rated 2 2mo agoA
tokens not measured
originalMIT
Multi-account Gmail + Google Calendar + Google Drive + Google Tasks + Contacts MCP server. Every tool takes an account slug (e.g. work | personal), so one Claude session can act across several Google accounts in parallel — send/draft/delete email, manage calendars, move/upload/share Drive files, manage tasks, and.
★not rated 2
changed 1mo agoA
tokens not measured
originalMIT
Skills for writing, sprint work, codebase workflows, and neurodivergent-friendly executive-function support (ink), plus a multi-lens decision council (agency).
Conduct a parallel multi-session sprint: one conductor, several work managers in git worktrees, a file-based mailbox, watchdog supervision, and a clean teardown.
★not rated 2 3mo agoA
tokens not measured
originalMIT
DNS, IP intelligence, AS routing, domain reputation, Bitcoin blockchain, Lightning Network, and threat intelligence. 50+ tools for network analysis, crypto investigation, and security research.
Compiler-aware C# navigation for .NET repos — Roslyn-backed symbol search, reference tracing, call and type hierarchies, and a solution-wide rename, with an index that follows every edit.
★not rated 2
changed 6d agoA
tokens not measured
originalMIT
Fact-checking skill for journalists — evaluates news articles using TTT/VIA (news classification) and AQTAQTA/IMVAIN (source quality) methodologies. Auto-detects Kazakh and English text, and detects genre (news, expert column, interview, press release) to adapt scoring.
★not rated 2 4mo agoA
tokens not measured
originalMIT
Google Workspace for Claude Code — Gmail, Calendar, Drive, Docs, Sheets, Forms, Tasks, Analytics (GA4), Search Console, Gmail Postmaster Tools and YouTube CLIs and skills, plus a gw api/gw schema escape hatch to any Google API. Multi-account OAuth with keyring-backed tokens.
★not rated 2
changed 13d agoA
tokens not measured
originalApache-2.0
A retro arcade in a tmux side-pane. Play Snake, 2048, Tic-Tac-Toe, Connect Four, and Chess while Claude Code works. Games run while Claude thinks and pause when it's done.
Spec & context-driven framework for Claude Code development. Structured skills for product planning, spec writing, task scoping, and context-aware implementation.
★not rated 2
changed 3d agoA
tokens not measured
originalMIT
A usage sensor, not a HUD. Claude Code exposes subscriber ratelimits (5-hour and 7-day) to the statusline, so the statusline feeds them to a Stop-hook sensor that, when you near a cap with a reset still far off, recommends or auto-triggers /genesis:close - a clean handoff instead of a hard cutoff. Optional mid-turn wa.
★not rated 2 5mo agoA
tokens not measured
originalMIT