{"id":"49967924-4ff2-4d1c-885c-7fe59aa36d2c","slug":"radin","kind":"skill","name":"radin","summary":"🐀 the agentic stack for those who have to save tokens","owner":"shortcuts","repoUrl":"https://github.com/shortcuts/radin","license":"MIT","licenseTier":"permissive","trust":"community","stars":6,"match":null,"fewStars":false,"format":"skill","createdAt":"2026-07-20T08:24:58.000Z","addedAt":"2026-10-14T05:59:06.320Z","logoUrl":"https://avatars.githubusercontent.com/u/20689156?u=b9559f176b8317f628ecb4560ef0677b7999a38e&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["developer-tools"],"tags":["claude-code","llm-tools","ai","claude-code-plugin","prompt-engineering"],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-25T19:40:08.526Z","sample":false,"url":"https://conduits.dev/tools/radin","language":"Shell","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"<p align=\"center\">\n  <strong>🐀 radin</strong>\n</p>\n\n<p align=\"center\">\n  <em>Too cheap to pay full price for a whole AI tool stack, so it went shopping for you.</em>\n</p>\n\n<p align=\"center\">\n  <sub>\"Radin\" is French slang for miser. Hence the rat.</sub>\n</p>\n\n<p align=\"center\">\n  <a href=\"LICENSE\"><img src=\"https://img.shields.io/badge/license-MIT-blue?style=flat\" alt=\"License\"></a>\n  <a href=\"AGENTS.md\"><img src=\"https://img.shields.io/badge/platform-macOS%20%7C%20Linux-lightgrey?style=flat\" alt=\"macOS | Linux\"></a>\n</p>\n\nA Claude Code plugin for a solo dev on a small subscription. One `curl | bash`\ngives you:\n\n- **A backlog-driven workflow.** Tasks live as files in your repo, so they\n  survive the conversation. Plan them, execute them one commit at a time,\n  review the result. Any run resumes where it stopped.\n- **A curated token-reduction stack**, installed whole: rtk, caveman,\n  ponytail, codebase-memory-mcp, thermo-nuclear, mattpocock-skills,\n  unslop, technical-writing.\n  radin delegates to them instead of reimplementing them. It never forks or\n  vendors them.\n\n## Install\n\n```sh\ncurl -fsSL https://raw.githubusercontent.com/shortcuts/radin/main/install.sh | bash\n```\n\nNeeds `curl`, `tar`, `bash`, and the `claude` CLI. Two files are C, the TUI and the\nJSON surgery behind `radin repair`. Install builds both with `cc` (Xcode\nCommand Line Tools on macOS, gcc on Linux) and skips each with a warning when\nno compiler is there. Companion tools pull their\nown stacks; one that fails is reported and skipped, and radin still installs.\n`rtk` installs through Homebrew, mise, or its own\ninstaller. The install asks which.\n\nInstall asks three questions: one about `radin-batch`'s sub-agent models,\none naming the package manager for companion tools, one about sub-agent effort. Add `-s -- --yes` to take the defaults. Add `-s -- --verbose` to\ntrace every command, for an install that hangs or fails.\n\nUpdate radin itself with `radin update`. It asks the sub-agent model and effort questions\nagain, reapplies every companion's configuration, and installs only the\ncompanions that are missing. Re-run `install.sh` to update the companions too.\n\n## Use it\n\n| Command | What it does |\n| --- | --- |\n| `/radin-record` | Turn a bug, idea, or aside raised mid-session into a backlog task |\n| `/radin-plan` | Write a step-by-step plan for one task, touching no code |\n| `/radin-batch` | Work the backlog: prioritize, plan the tasks `/radin-record` flagged, implement, commit, one task per commit |\n| `/radin-do` | Record one ad-hoc prompt as a backlog entry, plan it when flagged, then execute just that task |\n| `/radin-review` | Two-axis pass over a scope, repo rubrics plus the originating entry's spec; kept findings go to the backlog or straight to `/radin-do` |\n| `/radin-stats` | Every companion tool's own savings report, side by side |\n| `/radin-uninstall` | Remove what `install.sh` put in `~/.claude` |\n| `radin` | Full-screen backlog browser for humans. Bare `radin` on a terminal is the human entry point. No agent involved; `?` lists keys. Off a terminal it prints the usage text and exits non-zero, so a skill or pipe never hangs in a full-screen app. `radin help` prints that text on purpose |\n\nTypical loop: record, plan (optional), execute, review. Review findings feed the next\nexecute. Each skill runs in your own conversation, so it can ask you\nquestions; only leaf work goes to sub-agents. `/radin-batch` confirms the\norder before it starts, and every task's state is on disk, so interrupting it\ncosts nothing. Want your thread free? Dispatch it from `claude agents` or a\nsecond `claude` session. The worktree-per-task answer keeps concurrent runs\noff each other's checkout.\n\n## Where things live\n\nOne directory per repo, `<repo-root>/.claude/.radin/`, holds the backlog index, one\nmarkdown file per task, plans, reviews, and execution state. Commit it to\nshare the backlog with your team, or gitignore it to keep it private. radin\nnever touches your `.gitignore`.\n\nEvery tool goes through the `radin` CLI (`radin backlog`, `radin state`), so\nnothing hand-edits those files. A task carries a category\n(`feat`/`fix`/`chore`/`refactor`), an optional priority and `depends_on`, and\ncan sit under an epic whose `DESCRIPTION.md` every child inherits.\n\n## What it shops for\n\n| Tool | What it does |\n| --- | --- |\n| [rtk](https://github.com/rtk-ai/rtk) | Wraps dev commands, cuts their output 60-90% |\n| [caveman](https://github.com/JuliusBrussee/caveman) | Terse agent prose, ~65% fewer tokens |\n| [ponytail](https://github.com/DietrichGebert/ponytail) | Makes the agent think like the laziest senior dev in the room |\n| [codebase-memory-mcp](https://github.com/DeusData/codebase-memory-mcp) | Indexes the repo into a knowledge graph, so the agent asks instead of grepping |\n| [thermo-nuclear](https://github.com/cursor/plugins/tree/main/cursor-team-kit/skills/thermo-nuclear-code-quality-review) | Maintainability review skill |\n| [mattpocock-skills](https://github.com/mattpocock/skills) | Engineering skills; radin delegates interviewing and research here |\n| [unslop](https://github.com/cursor/plugins/tree/main/pstack/skills/unslop) | Rewrites prose to strip AI-slop patterns |\n| [technical-writing](https://github.com/cursor/plugins/tree/main/pstack/skills/technical-writing) | Technical-writing rules for docs and agent-facing prose |\n\nunslop and technical-writing are fetched from cursor/plugins `main` on every\ninstall and update, unpinned. An upstream edit changes agent behavior without\nreview.\n\nTwo caveats. A graph answer is a pointer, never proof of absence, so read the\nfile before editing. After `codebase-memory-mcp update`, run\n`radin repair`, because that update reruns a config write that\ndrops other tools' hooks. radin snapshots `~/.claude/settings.json` and\n`~/.claude.json` into `~/.claude/.radin/backups/` before that write and never\ndeletes a snapshot; `codebase-memory-mcp uninstall` undoes upstream's own side.\nFull list in\n[docs/DECISIONS.md](docs/DECISIONS.md).\n\nHacking on radin itself? [AGENTS.md](AGENTS.md),\n[CONTRIBUTING.md](CONTRIBUTING.md).\n"}