Skip to content
Back to search

rsc-harness

Your agent invents things because it has no memory, and can't touch your database because it has no arms. rsc is the meta-harness that gives it both, plus the trade to know the job and the reflexes to not break your repo. For Claude Code, Codex and 15 more.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026. Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/ericrisco/rsc-harness with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/ericrisco/rsc-harness'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/ericrisco--rsc-harness/badge.svg)](/listings/ericrisco--rsc-harness)

HTML

<a href="/listings/ericrisco--rsc-harness"><img src="/api/listings/ericrisco--rsc-harness/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

npx @ericrisco/rsc@latest onboard

Prefer the short `rsc` command? Install once, globally:

npm install -g @ericrisco/rsc   # then just: rsc

Run it inside any project and describe what you want. Working on the catalog itself? Clone and link:

git clone https://github.com/ericrisco/rsc-harness.git ~/rsc-skills
cd ~/rsc-skills && npm install && npm link

> **Run it inside the project you're equipping — not inside this repo.** The > catalog's own `package.json` is named `@ericrisco/rsc`, so `npx @ericrisco/rsc` > *from within a `rsc-harness` clone* resolves to the local (unlinked) bin and > dies with `sh: rsc: command not found`. Working on the catalog itself? Use > `node scripts/rsc.js …`, the `npm link` above, or pin the published build with > `npx @ericrisco/rsc@latest …`.

The first run asks **one** question about conversation: technical, or with analogies. Then what the project is for, its goal and the assistants to target. It then presents the complete plan. A small website can defer SDD, agents and code guards; an operations harness does not receive them merely because it lives in a repository. Deferred components record the evidence that would make rsc recommend them later. `rsc reassess` reports that evidence but still cannot install anything without a newly accepted plan.

Everything stays **in the project**, and the real skill files are written **once** to `.rsc/skills/<id>/`. Each assistant you pick gets a lightweight symlink back to that shared base — no copy is duplicated across IDEs. (If the filesystem can't symlink, it falls back to a real copy automatically.) On targets with file-based agents, the four base agents are installed too; stack specialists remain selective. Native command targets receive only entry points whose backing skill, agent or local-memory capability actually exists.

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 771 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <div align="center">
  2. <img src="https://raw.githubusercontent.com/ericrisco/rsc-harness/main/site/og.png" alt="rsc-harness: you build the product, rsc-harness builds your agent's harness, following strict best practices instead of improvising one." width="960">
  3. # `rsc-harness`
  4. **You build the product. rsc-harness builds its harness.**
  5. [![npm](https://img.shields.io/npm/v/@ericrisco/rsc?color=63d68a&labelColor=12161c&label=npm)](https://www.npmjs.com/package/@ericrisco/rsc)
  6. [![downloads](https://img.shields.io/npm/dm/@ericrisco/rsc?color=63d68a&labelColor=12161c&label=downloads)](https://www.npmjs.com/package/@ericrisco/rsc)
  7. [![skills](https://img.shields.io/badge/skills-273-63d68a?labelColor=12161c)](#the-catalog)
  8. [![license](https://img.shields.io/badge/license-MIT-63d68a?labelColor=12161c)](LICENSE)
  9. [![stars](https://img.shields.io/github/stars/ericrisco/rsc-harness?color=63d68a&labelColor=12161c)](https://github.com/ericrisco/rsc-harness/stargazers)
  10. Keep your project knowledge, skills and working rules together across the assistants you already use.
  11. Start with your goal. Review what rsc proposes to install. Accept the plan before it changes your project.
  12. **MIT licensed · project-local setup · 19 assistant targets**
  13. Give this URL to your agent and paste the request in the chat:
  14. ```text
  15. Read https://ericrisco.github.io/rsc-harness/ and set up rsc for this project.
  16. Ask me the onboarding questions, show me the exact plan, and wait for my acceptance before writing.
  17. ```
  18. Or start from the terminal:
  19. ```bash
  20. npx @ericrisco/rsc@latest onboard
  21. ```
  22. </div>
  23. ---
  24. ## Your project has a way of working. Give it to your agent.
  25. A new chat should not make you reconstruct the stack, explain yesterday's decision and remember which
  26. files must stay untouched. Putting every instruction into one giant prompt creates a different problem:
  27. the important details compete with rules that have nothing to do with today's task.
  28. Hand-built setups can become a **Frankenstein harness**: duplicated instructions, unrelated skills,
  29. premature hooks and unnecessary MCP connections. Open standards cover individual pieces
  30. ([Agent Skills](https://agentskills.io/specification), [MCP](https://modelcontextprotocol.io/),
  31. [AGENTS.md](https://agents.md/)). Your project still needs someone to choose and maintain the right pieces.
  32. rsc-harness makes that choice explicit. It reads evidence within your project root, asks about your goal
  33. and assistants, and prepares a plan you can inspect before applying it.
  34. ## See the plan before you install
  35. <img src="https://raw.githubusercontent.com/ericrisco/rsc-harness/main/site/meta-harness.png" alt="A project goal becomes an installation plan: components selected, deferred or excluded with reasons." width="960">
  36. Every component has a disposition:
  37. | In your plan | What it means |
  38. | --- | --- |
  39. | **Selected** | There is a reason to install it for this project now. |
  40. | **Deferred** | It may become useful; the plan records what evidence would justify adding it. |
  41. | **Excluded** | It does not belong in the proposed setup. |
  42. You accept the exact plan. If project evidence changes before acceptance, rsc asks you to review a
  43. fresh plan. Later, `rsc reassess` reports new needs without silently installing them.
  44. ### Why documentation comes first
  45. The decisions your agent needs should have a place to live outside the chat.
  46. | Part of the harness | What you get |
  47. | --- | --- |
  48. | **Knowledge** | A local `02-DOCS/` wiki for project context, decisions and evidence. |
  49. | **Tools** | A `01-TOOLS/` convention for documenting real connections and their checks. |
  50. | **Skills** | A selection from 273 available skills, loaded progressively when relevant. |
  51. | **Controls** | Working rules and provider-specific hooks, with explicit support limits. |
  52. The mechanism is practical: keep the decision, make it discoverable, and have the next task consult it.
  53. That gives the agent a maintained source to work from. It does not make an outdated note correct or
  54. remove the need to check the result.
  55. ## A day with rsc
  56. <img src="https://raw.githubusercontent.com/ericrisco/rsc-harness/main/site/lanes.svg" alt="The agent chooses a read-only answer, a focused change with evidence, or a specification-led workflow for coupled decisions." width="960">
  57. 1. **Describe the project.** Choose the assistants you use and how technical the conversation should be.
  58. 2. **Review the setup.** See each proposed component and its reason, then accept the plan you want applied.
  59. 3. **Ask for the work.** The agent names the appropriate lane: an answer, a focused change (FTD), or the
  60. specification-led chain (ASDD) for decisions that affect each other.
  61. 4. **Keep the result and its evidence.** Project knowledge goes into `02-DOCS/`; implementation stays in
  62. your code. Future sessions can consult both.
  63. ## Is this for your project?
  64. Use rsc when you already work with an assistant and want a coherent project setup you can inspect,
  65. share and maintain. It can equip software projects and business workflows; the catalog includes both.
  66. You keep your editor, model and existing project. rsc supplies the harness around them. Provider
  67. features differ: check [supported assistants](#supported-assistants) before relying on a particular
  68. hook or memory feature. Model and external-service costs remain with their providers.
  69. For a one-off question, a maintained project harness may be unnecessary. You can also install an
  70. individual skill into an existing rsc project rather than adopt a whole workflow.
  71. ## Install in one minute
  72. ```bash
  73. npx @ericrisco/rsc@latest onboard
  74. ```
  75. The first run asks **one** question about conversation (technical, or with analogies), then what the
  76. project is for, its goal and the assistants to target. Then it presents the complete plan. A small
  77. website can defer ASDD, agents and code guards; an operations harness does not receive them merely
  78. because it lives in a repository. Deferred components record the evidence that would make rsc
  79. recommend them later, and `rsc reassess` reports it without installing anything.
  80. Everything stays **in the project**. The real skill files are written once to `.rsc/skills/<id>/`, and
  81. each assistant gets a symlink back to that shared base, so picking several duplicates nothing. Skills
  82. load progressively: only their description is paid for until a task needs them.
  83. Prefer a short `rsc` command? `npm install -g @ericrisco/rsc`.
  84. When an executable `rsc` is already on PATH, install and sync write operational skill and
  85. command instructions using that installed CLI. This also applies to unversioned commands
  86. suggested by hooks. Explicit versions such as `@4.1.1` and `@latest` keep their meaning.
  87. No execution permission is added for this choice. If the CLI is removed, run sync using the
  88. version declared in `.rsc.json` to restore the `npx` instructions.
  89. > **Run it inside the project you are equipping, not inside a clone of this repo.** The catalog's own
  90. > `package.json` is named `@ericrisco/rsc`, so `npx @ericrisco/rsc` from within a clone resolves to
  91. > the local bin and fails with `sh: rsc: command not found`. Working on the catalog itself? Use
  92. > `node scripts/rsc.js …` or `npm link`.
  93. ### For agents: "install this harness"
  94. If you are an AI coding agent and your user asked you to **install rsc, this harness or these
  95. skills**, run this in the root of the project they are equipping:
  96. ```bash
  97. npx @ericrisco/rsc@latest onboard
  98. ```
  99. It asks the user how technical to talk, the project kind and the goal **before it writes anything**,
  100. and shows every selected or deferred skill, agent and hook with its reason. Relay those questions;
  101. never invent or default the answers. The first pass writes nothing and prints a SHA-256 plan id. Rerun
  102. with `--accept-plan <id>` only after the user accepts that exact plan. If the project already has a
  103. skill of its own with a catalog name, the plan lists it under **Your own skills this plan replaces**:
  104. tell the user before accepting.
  105. Non-interactive, with the answers collected first:
  106. ```bash
  107. npx @ericrisco/rsc@latest onboard --technical-level technical \
  108. --project-kind software --goal "small compound-interest website" \
  109. --software-scope small --target codex
  110. # After the user accepts the printed plan:
  111. npx @ericrisco/rsc@latest onboard --technical-level technical \
  112. --project-kind software --goal "small compound-interest website" \
  113. --software-scope small --target codex --accept-plan PRINTED_SHA256_ID
  114. ```
  115. If project evidence changes between preview and acceptance, rsc returns `RSC_PLAN_CHANGED` and writes
  116. nothing. After applying, it verifies the result and prints `RSC_ONBOARDING_READY`, or
  117. `RSC_ONBOARDING_INCOMPLETE` with each missing path and the action that creates it.
  118. ---
  119. ## How it works inside
  120. ### The knowledge model: `02-DOCS/` and `01-TOOLS/`
  121. <img src="https://raw.githubusercontent.com/ericrisco/rsc-harness/main/site/company-brain.svg" alt="The rsc company brain as a 3D knowledge graph: loose material drifts in from the inbox and is absorbed into the wiki cluster, whose brightest hubs are its .base views; dim clusters are raw sources and the agent's own worklog; an amber cluster is gaps.md, what the wiki knows it is missing." width="960">
  122. Anything you drop into `02-DOCS/inbox/` is ingested. The wiki consolidates it as plain markdown that
  123. every skill reads before acting, so the next session starts where the last one stopped. Raw sources
  124. and the agent's worklog stay underneath, never deleted. You can open `02-DOCS/` as an Obsidian vault,
  125. and every `02-DOCS/wiki/` is a valid [Open Knowledge Format](https://github.com/GoogleCloudPlatform/knowledge-catalog/tree/main/okf)
  126. bundle: markdown with YAML frontmatter, standard links, and the reserved `index.md` and `log.md`.
  127. `01-TOOLS/` holds one folder per tool: `.env.example`, `CREDENTIALS.md`, a README and a
  128. `test_connection` that proves it works. When you name a tool ("connect my Holded", "can the agent
  129. read our Sage?"), the `connect-tool` skill researches the vendor's current docs, picks the least
  130. dangerous route (official MCP, API, export, a read-only copy of the database), and builds the folder.
  131. For a read-only connection, the test also proves that writing fails.
  132. ### Three lanes, and the agent picks
  133. Every turn takes exactly one lane, and the agent names it in one line, in your language.
  134. 1. **Answer.** The request asks for information: explain, compare, investigate, audit, review,
  135. propose. It is **read-only**: nothing is written and no artifact is created. When intent is
  136. unclear it asks one question and stays here.
  137. 2. **FTD, Fast-Track Development.** The request authorises a change. This is the default for ordinary
  138. work. One feature document holds intent, scope, a checklist, the evidence and the next step. Tasks
  139. are checked off against observed proof, never against intention.
  140. 3. **ASDD, Agent Spec-Driven Development.** For the big or complex: several decisions that affect
  141. each other, and work an agent may run for hours. **The agent enters it on its own**, and you still
  142. decide *what* gets built: the spec answers four questions (what the user can do, what result they
  143. get, what it must respect, how it proves it) and you approve the spec. Then every task is reviewed inside
  144. the build, scaled by `rsc risk`; what the agent must not decide alone waits in a decision queue
  145. while independent work continues. Low-risk work lands on green CI; auth, billing, CI and migrations
  146. wait for you.
  147. Ask for the other lane and it switches.
  148. ### Skills, and how they are recommended
  149. The unit of installation is **one skill**: install `fastapi` without ever pulling `go`. Two faces
  150. share one catalog (`manifest.json`):
  151. - **In the terminal**, `rsc consult` ranks the catalog against your words and what it detects in the
  152. repo (`package.json` + `next` → `nextjs`, `go.mod` → `go`, `Dockerfile` → `docker`).
  153. - **In the chat**, the always-on `suggest` notices when a task needs a skill you do not have, names
  154. it, and installs it on a one-word yes.
  155. Skill authoring follows a research → spec → implement → review process. The public
  156. [rubric](scripts/skill-rubric.md) separates document quality from
  157. [behavioral evaluation](scripts/skill-behavior-rubric.md). Structural checks validate the catalog;
  158. behavioral comparisons ask whether loading a skill improves the result. Neither establishes a
  159. guaranteed business outcome.
  160. ### Hooks and guards
  161. In Claude Code, hooks enforce what instructions only ask for: the lane decisor on every turn, a
  162. refusal before a commit on a closed `main`, a confirmation before destructive commands, and a nudge
  163. to write the feature document before the first code edit. Every refusal says how to carry on. The
  164. rules only see what the **agent** runs; a person committing in their own terminal is never touched.
  165. > **Limit, stated plainly.** Those guards run in **Claude Code**. In Codex, Gemini, Cursor and
  166. > OpenCode the same rules travel in the skills and the session memory. They are followed, but no hook
  167. > enforces them.
  168. ### Working as a team
  169. Day ninety is where harnesses break, especially with several people, assistants and sessions on one
  170. project. rsc handles four things by default:
  171. | | What happens | Turn it off |
  172. | --- | --- | --- |
  173. | **You choose how the project works** | The install asks once: straight on `main`, or branches and pull requests? With branches, a commit on `main` is refused and the agent asks before each change: this branch, a new one, or unlock `main`? It never opens a branch on its own. | «unlock main», or `rsc main unlock` |
  174. | **One workspace per agent** | If another session works in the same folder, new work goes to `.worktrees/<branch>/` at the root you work in (never inside a project folder) and is never committed. A `post-merge` hook retires it once the branch lands. | `rsc isolation off` |
  175. | **The agent picks the method** | FTD for simple changes, ASDD for complex ones. The agent decides and says why. | Ask for the other lane |
  176. | **Knowledge reaches everyone** | `01-TOOLS/`, `02-DOCS/wiki/` and `02-DOCS/attachments/` travel through one exchange branch, `rsc/knowledge`, and reach `main` inside your normal pull requests. | `rsc knowledge-sync off` |
  177. **Emoji.** The install asks once whether rsc may use emoji (`--emoji <on|off>`, default on). With it off, rsc writes `.rsc/.no-emoji` and uses none: not in replies, not in commits, not in its own output. Change it later with `rsc emoji <on|off>`; `rsc doctor` reports it.
  178. **Orca worktrees.** Set `"worktreeManager": "orca"` in `.rsc.json` to retain externally managed
  179. checkouts even when Orca is unavailable. rsc also recognises a matching Orca CLI response, and never
  180. removes these checkouts through plain Git. [Setup and cleanup](skills/worktrees/references/orca.md)
  181. use the pinned harness version and Orca's lifecycle hooks; workspace lineage does not silence
  182. warnings about unverified concurrent sessions.
  183. **How it knows a project is complex.** A hook counts what anyone can check: a CI setup, a deployment
  184. file (`Dockerfile`, `vercel.json`, `fly.toml`…) or at least two people in the last 50 commits. The
  185. answer you gave at install always wins over this detection.
  186. #### Knowledge sync, step by step
  187. ```text
  188. ana · feat/login edits 02-DOCS/wiki/api.md ──► rsc/knowledge (when her turn ends)
  189. eric · feat/payments 📥 1 change from Ana ◄── rsc/knowledge (before his next message)
  190. main · protected untouched, gets it inside the next merged pull request
  191. new · clones main has it all after the first message
  192. ```
  193. - **When a turn ends**, your changes in those folders are committed on your branch as
  194. `📝 docs(auto): …` and sent to `rsc/knowledge`. Only the copy on `rsc/knowledge` carries
  195. `[skip ci]`, so your own pushes still run CI. On a closed `main`, or with code of yours still
  196. uncommitted, nothing is committed on your branch: the snapshot only goes to `rsc/knowledge`.
  197. - **Before each message**, what teammates sent comes into the branch you are on as a
  198. `📥 docs(auto): sync` commit, and you are told in one line. Only knowledge folders are touched.
  199. - **It never** pushes to `main`, pushes your code, brings in anybody else's code, syncs your personal
  200. `user-profile.md`, or overwrites a file you are editing.
  201. It runs in the turn hooks of Claude Code, Codex, Gemini, Cursor, OpenCode and DeepSeek Harness. It is
  202. a **project** switch: if one person stopped sending, the rest would stop seeing their work.
  203. #### Sharing the harness through git
  204. **Commit these:** `.rsc.json` (which assistants, which skills, which catalog version), `01-TOOLS/`,
  205. `02-DOCS/`, and skills or agents you wrote by hand. **rsc ignores the rest for you:** `.rsc/` (machine
  206. state) and the skill entries it manages (symlinks here, copies on Windows).
  207. Three files carry the harness through git: `.rsc.json`, `.claude/settings.json` and
  208. `.claude/rsc-bootstrap.mjs`, the small file that notices in a clone that the rest is not built yet.
  209. Commit all three. Because `rsc-bootstrap.mjs` runs on every session, review changes to it the way you
  210. would review a CI workflow.
  211. Whoever clones runs **one command** and gets the same harness, at the version the project pinned in
  212. its `.rsc.json` as `catalogVersion`:
  213. ```bash
  214. npx @ericrisco/rsc@<catalogVersion> sync
  215. ```
  216. You do not have to know that. Open the project and the assistant tells you, in the first message,
  217. what is missing and the exact command, naming the pinned version. It asks; it never installs on its
  218. own. A teammate who clones in three months gets what you had, not what shipped since. After a
  219. `git pull`, `rsc doctor` tells you what no longer matches; nothing is written by a pull.
  220. **Own skills.** A skill your team wrote lives in the repo and works for whoever clones. rsc never installs, updates or overwrites
  221. it: its version is the commit. Installing a catalog skill with the
  222. same name is refused, and `doctor` lists what is yours by reading the files. Declaring a skill under
  223. `ownSkills` still works.
  224. #### Git permissions
  225. A harness installed from scratch lets the agent run `git commit`, `git push` and `gh pr create`
  226. without asking, because those steps close every lane. A force-push still asks, and in Claude Code a
  227. commit on a closed `main` is still refused. It is saved as `gitPermissions` in `.rsc.json`.
  228. | Assistant | Where it goes |
  229. | --- | --- |
  230. | Claude Code | `.claude/settings.json` → `permissions.allow` (force-push under `ask`) |
  231. | Codex | `.codex/rules/rsc-git.rules` (force-push → `prompt`) |
  232. | Gemini | `.gemini/settings.json` → `tools.allowed` |
  233. | OpenCode | `opencode.json`: the `permissions` list on OpenCode 2, `permission.bash` on 1.x. An `opencode.jsonc` is never rewritten; rsc prints the rules to paste. |
  234. | Cursor, DeepSeek Harness | Not covered: neither has a safe per-command allow list. |
  235. `rsc git-permissions status | on | off` explains, per assistant, what rsc controls and what is left
  236. to the assistant.
  237. ### Memory between sessions
  238. On supported local targets, rsc checkpoints observable repository state (branch, worktree, HEAD,
  239. changed paths, commits, SDD status) and injects it before the first action of a new session in the
  240. same checkout.
  241. - **Full:** Claude Code, Codex, Gemini CLI, OpenCode and DeepSeek Harness. Codex asks you to trust the
  242. project hook once with `/hooks`.
  243. - **Assisted:** Cursor desktop, through an always-on rule that reads before acting.
  244. - **Never cloud:** the memory runtime makes no network request.
  245. The journal never stores prompts, responses, tool output, file contents or secrets. It stays in a
  246. git-excluded local path, keeps 30 days, and injects at most 4,096 bytes. `rsc memory off` disables it.
  247. ### Review that only brings you what is real
  248. In ASDD, review runs inside every task, as deep as `rsc risk` says: none for docs, one fresh-context
  249. reviewer for ordinary code, three adversarial reviewers (correctness, security, tests) for auth,
  250. billing, CI or migrations. `rsc review ledger` caps the fix loop at three rounds and never lets a
  251. finding be raised twice, and `rsc review escapes` turns every revert into a case for the reviewer. The
  252. three reviewers
  253. often report the same defect in different words, and a reviewer hunting for problems can overstate
  254. one. So `rsc review consolidate` merges the duplicates, and every serious finding left goes to a
  255. fresh `finding-verifier` agent whose starting position is that the finding is false. It blocks the
  256. merge only if the verifier reproduces or traces it. You get "11 reported → 4 unique → 2 confirmed",
  257. with the refuted ones listed and their reason.
  258. ### It talks so you understand
  259. - **Short sentences, one idea each,** in the style of ASD-STE100 (the controlled writing of aircraft
  260. manuals), always in your language.
  261. - **The answer first.** Every answer stands alone, without "that fix" or "as above".
  262. - **One setting:** technical, or with analogies.
  263. - **A ladder when you do not follow:** text, then a diagram, then an HTML page, and a video only if
  264. you ask.
  265. Every turn closes with a short compass: where you are, and the next step as a question. Text you send
  266. to other people (emails, posts, READMEs) goes through `unslop`, so it sounds like a person.
  267. ---
  268. ## Supported assistants
  269. The real files land once in `.rsc/skills/<id>/`; each assistant gets a symlink or a converted file.
  270. The wizard asks which ones; `--target a,b` does it non-interactively.
  271. | Target | Skill destination (→ `.rsc/skills/<id>/`) | Always-on detector |
  272. | --- | --- | --- |
  273. | `claude` | `.claude/skills/<id>/` → symlink (copy on Windows) | SessionStart hook in `.claude/settings.json` |
  274. | `codex` | `.codex/rsc/<id>/` → symlink | block in `AGENTS.md` |
  275. | `copilot` | `.github/rsc/<id>/` → symlink | block in `.github/copilot-instructions.md` |
  276. | `cursor` | `.cursor/rules/<id>.mdc` (converted) | always-apply rule |
  277. | `gemini` | `.gemini/rsc/<id>/` → symlink | block in `GEMINI.md` |
  278. | `windsurf` | `.windsurf/rsc/<id>/` → symlink | rule in `.windsurf/rules/rsc-suggest.md` |
  279. | `cline` | `.clinerules/rsc/<id>/` → symlink | rule in `.clinerules/rsc-suggest.md` |
  280. | `antigravity` | `.antigravity/rsc/<id>/` → symlink | block in `.antigravity/AGENTS.md` |
  281. | `zed` | `.zed/rsc/<id>/` → symlink | block in `AGENTS.md` |
  282. | `continue` | `.continue/rsc/<id>/` → symlink | rule in `.continue/rules/rsc-suggest.md` |
  283. | `roo` | `.roo/rsc/<id>/` → symlink | rule in `.roo/rules/rsc-suggest.md` |
  284. | `amp` | `.amp/rsc/<id>/` → symlink | block in `AGENTS.md` |
  285. | `opencode` | `.opencode/rsc/<id>/` → symlink | block in `AGENTS.md` |
  286. | `deepseek` | `.dsh/skills/<id>/` → symlink (discovered natively) | block in `AGENTS.md` |
  287. | `pi` | `.pi/skills/<id>/` → symlink (copy on Windows; discovered natively) | block in Pi's selected project context file |
  288. | `jules` | `.jules/rsc/<id>/` → symlink | block in `AGENTS.md` |
  289. | `junie` | `.junie/rsc/<id>/` → symlink | block in `.junie/guidelines.md` |
  290. | `kiro` | `.kiro/rsc/<id>/` → symlink | doc in `.kiro/steering/rsc-suggest.md` |
  291. | `aider` | `.aider/rsc/<id>/` → symlink | block in `CONVENTIONS.md` |
  292. > `codex`, `zed`, `amp`, `opencode`, `deepseek` and `jules` share one root `AGENTS.md`; the block is
  293. > written once. Pi shares it when that is the context file its native loader selects.
  294. > **DeepSeek Harness** (`dsh`) has no project-level hooks, so once per machine rsc adds one marked
  295. > block to `~/.dsh/cordis.patch.yml` that runs the hooks of whichever project the session is in.
  296. > Restart dsh once after the first install; delete the block to turn it off.
  297. > **Pi:** select `--target pi` during onboarding, then run `/reload` in Pi to discover the
  298. > installed skills. Invoke one explicitly with `/skill:<id>`. Pi loads project resources after
  299. > project trust is granted; rsc preserves your `.pi/settings.json` and extensions.
  300. > This target uses Pi's [native skills and context files](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/docs/configuration.md).
  301. > Its [context loader](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/src/core/resource-loader.ts)
  302. > selects the first readable file in this order: `AGENTS.override.md`, `AGENTS.md`, `AGENTS.MD`,
  303. > `CLAUDE.md`, `CLAUDE.MD`. rsc adds its marked block to that file, preserving your instructions;
  304. > it creates `AGENTS.md` only when none exists. If a new override hides the block, doctor reports
  305. > the missing wiring and sync writes the block into the selected file.
  306. > It does not install a custom extension, subagents, automatic session-memory hooks, or tool guards.
  307. > Pi has no built-in subagents; skills that require independent reviewers need an external review.
  308. > The instructions request an update check on the first turn, as for other targets without a wired session hook.
  309. | Targets | Stack agents | Native commands | Session memory |
  310. | --- | --- | --- | --- |
  311. | Claude Code | yes | agent + memory entries | full |
  312. | Codex | yes | no separate surface | full after `/hooks` trust |
  313. | Cursor desktop | yes | yes | assisted |
  314. | Gemini CLI, OpenCode | yes | yes | full |
  315. | DeepSeek Harness | unsupported | unsupported | full (machine bridge) |
  316. | Pi | unsupported | native `/skill:<id>` (no generated command files) | unsupported |
  317. | GitHub Copilot | yes | yes | unsupported |
  318. | Junie, Kiro | yes | unsupported | unsupported |
  319. | Windsurf, Cline, Roo | unsupported | yes | unsupported |
  320. | Antigravity, Zed, Continue, Amp, Jules, Aider | unsupported | unsupported | unsupported |
  321. `manifest.json` is the generated public inventory: 34 agents (5 base + 29 selective specialists) and
  322. 53 command entries (20 fixed + 33 stack aliases). Unsupported means rsc writes nothing for that
  323. surface. Generated agents carry no `model:` on OpenCode, so they use the session's model; pin one for
  324. the team with `rsc agent-model <target> <model>`. If you edit a generated agent, `rsc sync` keeps your
  325. file, and `rsc agents reset <name>` takes rsc's version back after a backup.
  326. ---
  327. ## CLI
  328. `rsc` below is short for `npx @ericrisco/rsc`. Without a global install (`npm i -g @ericrisco/rsc`),
  329. the bare `rsc` answers `command not found`: type the `npx` form.
  330. ```bash
  331. # Setup
  332. rsc onboard # plain-language onboarding; writes nothing until you accept
  333. rsc reassess # check deferred components against new evidence
  334. rsc add fastapi postgresdb # install skills by name
  335. rsc consult "I want to launch a SaaS" # recommend only
  336. rsc sync # refresh managed skills and hooks
  337. # Team
  338. rsc main status | unlock | lock # whether the agent may work on main
  339. rsc isolation status | off # worktrees for parallel sessions
  340. rsc knowledge-sync status | off # the shared knowledge exchange
  341. rsc git-permissions status | on | off
  342. # Agents and review
  343. rsc agent-model opencode inherit # agents use the session's model
  344. rsc agents status | reset <name> # installed agents, and the ones you edited
  345. rsc review consolidate c.md s.md t.md # merge reviewer findings; serious ones go to finding-verifier
  346. # Documentation
  347. rsc ingest notas.md --topic crm # move a file into 02-DOCS/raw/ (verified copy), then document it
  348. rsc copy rules | check ad.txt # the phrases this project forbids, and a check before publishing
  349. # Memory and diagnostics
  350. rsc memory status | off | resume # session continuity; «recuerda esto» goes to 02-DOCS/wiki/
  351. rsc doctor # harness health and onboarding readiness
  352. rsc repair # put the harness back to what is declared
  353. rsc backups | restore latest # project-local snapshots
  354. rsc uninstall <id> | purge # remove one skill, or everything rsc installed
  355. ```
  356. ### `rsc doctor`
  357. `doctor` answers two questions, on separate lines:
  358. ```text
  359. Harness health: healthy
  360. Onboarding readiness: pending
  361. Pending: 02-DOCS/wiki/sdd/constitution.md (draft)
  362. ```
  363. **Harness health** says whether what is installed works, and it is the only thing the exit code
  364. follows. **Onboarding readiness** says whether onboarding is finished (`ready`, `pending`,
  365. `incomplete`, or `not onboarded` for a manual install); it never changes the exit code.
  366. ### Something's off? `rsc repair`
  367. If the harness is wired for the wrong assistant, a hook runs several times, or skills appear that you
  368. never asked for, run `npx @ericrisco/rsc@latest repair`. It shows what it found before touching
  369. anything, keeps a recoverable copy, and never touches skills or agents you wrote by hand. Add
  370. `--dry-run` to see the pass without writing.
  371. ### Update
  372. **rsc updates itself.** At session start it checks npm. A release in the same major installs in the
  373. background and takes effect next session; a new major asks first. To turn it off, create
  374. `.rsc/.no-auto-update`. By hand: `npm install -g @ericrisco/rsc@latest && rsc sync`. Every sync
  375. snapshots the project first, so `rsc restore latest` undoes a bad update.
  376. ---
  377. ## Build a sales page and VSL from the same offer
  378. The **VSL sales pack**, added in 3.7.0, connects buyer evidence to the offer, the buying journey,
  379. and a video sales letter (VSL). It is a curated selection of ordinary rsc skills; each remains usable alone.
  380. | Step | Skill | Deliverable |
  381. | --- | --- | --- |
  382. | Understand the buyer | [customer-research](skills/customer-research/) | Sourced needs, objections and buyer language, with gaps and publication permissions. |
  383. | Define the exchange | [offer-design](skills/offer-design/) | What the buyer gets, why it helps, delivery limits and supported terms. |
  384. | Plan the route | [sales-funnel](skills/sales-funnel/) | Pages, decisions, exits, recovery paths and measurement handoffs. |
  385. | Write the video | [vsl-script](skills/vsl-script/) | Spoken script, feasible timing, scene cues and evidence for each claim. |
  386. | Write the page | [landing-copy](skills/landing-copy/) | Page text aligned with the same offer revision and video. |
  387. In an existing rsc project, this command adds the core selection:
  388. ```bash
  389. npx @ericrisco/rsc add customer-research offer-design sales-funnel vsl-script landing-copy
  390. ```
  391. Use `design` and your web-stack skill to build the page; use `remotion-video` or your editor to produce
  392. the video. Add `analytics` and `ab-testing` when you are ready to measure the result. These are separate
  393. capabilities, not an automatic payment integration or a promise that the page will convert.
  394. Try this request after installing the selection:
  395. ```text
  396. Use the research and product facts we already have to define one offer.
  397. Plan the shortest buying journey, write its VSL, and prepare the landing copy.
  398. Keep the price, scope, evidence and next action consistent across all three.
  399. Separate missing proof from publishable claims. Show what still needs production or implementation.
  400. ```
  401. **Evaluation status:** initial blind comparisons met the quality threshold for all four new skills;
  402. only `offer-design` met the required improvement over the no-skill baseline. The others ship with
  403. that limitation recorded in their `evals/README.md`. Repeated-trial validation remains open.
  404. The same workflow works for a free tool: the next action can be trying or installing it. No invented
  405. price, customer testimonial, deadline or guarantee is needed to explain a useful product.
  406. ---
  407. ## The catalog
  408. 273 skills, grouped by what you are trying to do. Click any skill to read its `SKILL.md`. It fires on
  409. its own when a task matches.
  410. ### 🧭 Core & control plane
  411. [init](skills/init/) · [harness](skills/harness/) · [orient](skills/orient/) · [suggest](skills/suggest/) · [unslop](skills/unslop/) · [author-skill](skills/author-skill/) · [sdd-init](skills/sdd-init/)
  412. ### 📐 Spec-Driven Development
  413. Two lanes for change: `ftd` for ordinary work, and the ten-phase chain when durable artifacts would
  414. settle a real ambiguity. `idea-refinement` runs inside `specify`; `decision-challenge` is on demand.
  415. [ftd](skills/ftd/) · [asdd](skills/asdd/) · [constitution](skills/constitution/) · [idea-refinement](skills/idea-refinement/) · [specify](skills/specify/) · [clarify](skills/clarify/) · [plan](skills/plan/) · [tasks](skills/tasks/) · [analyze](skills/analyze/) · [decision-challenge](skills/decision-challenge/) · [implement](skills/implement/) · [source-grounded-development](skills/source-grounded-development/) · [verify](skills/verify/) · [review](skills/review/) · [simplify-code](skills/simplify-code/) · [ship](skills/ship/) · [debug](skills/debug/) · [worktrees](skills/worktrees/) · [parallel](skills/parallel/) · [tidy](skills/tidy/)
  416. ### 💼 Run a business
  417. [finance-ops](skills/finance-ops/) · [invoicing](skills/invoicing/) · [bookkeeping](skills/bookkeeping/) · [pricing](skills/pricing/) · [sales-pipeline](skills/sales-pipeline/) · [lead-gen](skills/lead-gen/) · [cold-outreach](skills/cold-outreach/) · [proposals](skills/proposals/) · [contracts](skills/contracts/) · [customer-support](skills/customer-support/) · [client-onboarding](skills/client-onboarding/) · [retention](skills/retention/) · [hiring](skills/hiring/) · [people-ops](skills/people-ops/) · [inventory](skills/inventory/) · [logistics-ops](skills/logistics-ops/) · [procurement](skills/procurement/) · [meeting-notes](skills/meeting-notes/) · [sop-builder](skills/sop-builder/) · [project-ops](skills/project-ops/)
  418. ### 💸 Raise & model money
  419. [pitch-deck](skills/pitch-deck/) · [investor-materials](skills/investor-materials/) · [financial-model](skills/financial-model/) · [fundraising](skills/fundraising/) · [unit-economics](skills/unit-economics/) · [grants](skills/grants/)
  420. ### ⚖️ Legal, privacy & compliance
  421. [gdpr-privacy](skills/gdpr-privacy/) · [terms-conditions](skills/terms-conditions/) · [compliance](skills/compliance/) · [data-policy](skills/data-policy/) · [ip-trademark](skills/ip-trademark/)
  422. ### 📣 Market & brand
  423. [marketing](skills/marketing/) · [customer-research](skills/customer-research/) · [offer-design](skills/offer-design/) · [sales-funnel](skills/sales-funnel/) · [vsl-script](skills/vsl-script/) · [seo-geo](skills/seo-geo/) · [content-engine](skills/content-engine/) · [social-publisher](skills/social-publisher/) · [brand-voice](skills/brand-voice/) · [brand-identity](skills/brand-identity/) · [newsletter](skills/newsletter/) · [landing-copy](skills/landing-copy/) · [ads](skills/ads/) · [article-writing](skills/article-writing/) · [case-studies](skills/case-studies/) · [video-shorts](skills/video-shorts/) · [podcast](skills/podcast/) · [market-research](skills/market-research/) · [competitor-watch](skills/competitor-watch/) · [press-kit](skills/press-kit/) · [community](skills/community/) · [webinar](skills/webinar/) · [review-management](skills/review-management/)
  424. ### 🎬 Grow a channel
  425. Each with a `02-DOCS` feedback loop that learns from your own results.
  426. [youtube-api](skills/youtube-api/) · [youtube-strategy](skills/youtube-strategy/) · [youtube-ideation](skills/youtube-ideation/) · [youtube-thumbnails](skills/youtube-thumbnails/) · [youtube-packaging](skills/youtube-packaging/) · [remotion-video](skills/remotion-video/) · [tiktok-api](skills/tiktok-api/) · [instagram-api](skills/instagram-api/) · [shortform-strategy](skills/shortform-strategy/) · [shortform-ideation](skills/shortform-ideation/) · [shortform-packaging](skills/shortform-packaging/) · [shortform-editing](skills/shortform-editing/) · [viral-score](skills/viral-score/) · [linkedin-api](skills/linkedin-api/) · [linkedin-strategy](skills/linkedin-strategy/) · [linkedin-content](skills/linkedin-content/) · [linkedin-carousels](skills/linkedin-carousels/) · [linkedin-outreach](skills/linkedin-outreach/) · [medium-writing](skills/medium-writing/) · [medium-publishing](skills/medium-publishing/) · [medium-strategy](skills/medium-strategy/)
  427. ### 🔌 Connect & automate
  428. [connect-tool](skills/connect-tool/) · [stripe](skills/stripe/) · [email-connector](skills/email-connector/) · [google-workspace](skills/google-workspace/) · [notion-connector](skills/notion-connector/) · [whatsapp-telegram](skills/whatsapp-telegram/) · [automation-flows](skills/automation-flows/) · [api-connector-builder](skills/api-connector-builder/) · [webhooks](skills/webhooks/) · [data-scraper](skills/data-scraper/) · [spreadsheet-ops](skills/spreadsheet-ops/) · [calendar-scheduling](skills/calendar-scheduling/) · [document-processing](skills/document-processing/) · [e-signature](skills/e-signature/)
  429. ### ⚙️ Automation
  430. Drive the big automation platforms through their REST API or MCP server. `automation-strategy` decides
  431. whether, what and which platform.
  432. [automation-strategy](skills/automation-strategy/) · [n8n](skills/n8n/) · [make](skills/make/) · [zapier](skills/zapier/) · [power-automate](skills/power-automate/)
  433. ### 📊 Data & analytics
  434. [analytics](skills/analytics/) · [dashboard](skills/dashboard/) · [kpi-framework](skills/kpi-framework/) · [reporting](skills/reporting/) · [ab-testing](skills/ab-testing/) · [forecasting](skills/forecasting/) · [data-cleaning](skills/data-cleaning/) · [business-intelligence](skills/business-intelligence/)
  435. ### 🤖 AI: build it in
  436. [building-agents](skills/building-agents/) · [rag](skills/rag/) · [embeddings-search](skills/embeddings-search/) · [prompt-engineering](skills/prompt-engineering/) · [llm-pipeline](skills/llm-pipeline/) · [agent-eval](skills/agent-eval/) · [chatbot](skills/chatbot/) · [ai-media](skills/ai-media/) · [replicate-images](skills/replicate-images/) · [structured-extraction](skills/structured-extraction/) · [agent-safety](skills/agent-safety/) · [cost-tracking](skills/cost-tracking/)
  437. ### 🛰️ AI: run it on
  438. [replicate](skills/replicate/) · [runpod](skills/runpod/) · [modal](skills/modal/) · [huggingface](skills/huggingface/) · [ollama](skills/ollama/) · [together-fireworks](skills/together-fireworks/) · [fal](skills/fal/)
  439. ### 🎓 AI: train it
  440. [machine-learning](skills/machine-learning/) · [deep-learning](skills/deep-learning/) · [nlp](skills/nlp/) · [finetuning](skills/finetuning/) · [training-data](skills/training-data/) · [unsloth](skills/unsloth/) · [open-weights](skills/open-weights/) · [vllm](skills/vllm/)
  441. ### 🗣️ Languages
  442. [typescript](skills/typescript/) · [python](skills/python/) · [java](skills/java/) · [csharp-dotnet](skills/csharp-dotnet/) · [php](skills/php/) · [ruby](skills/ruby/) · [cpp](skills/cpp/) · [elixir](skills/elixir/) · [bash-scripting](skills/bash-scripting/) · [sql](skills/sql/) · [go](skills/go/)
  443. ### 🏗️ Frameworks & app stacks
  444. [fastapi](skills/fastapi/) · [nextjs](skills/nextjs/) · [react](skills/react/) · [react-native](skills/react-native/) · [vue-nuxt](skills/vue-nuxt/) · [angular](skills/angular/) · [svelte](skills/svelte/) · [astro](skills/astro/) · [solid-js](skills/solid-js/) · [htmx](skills/htmx/) · [nodejs](skills/nodejs/) · [nestjs](skills/nestjs/) · [django](skills/django/) · [laravel](skills/laravel/) · [rails](skills/rails/) · [spring-boot](skills/spring-boot/) · [phoenix](skills/phoenix/) · [flutter](skills/flutter/) · [swift-ios](skills/swift-ios/) · [kotlin-android](skills/kotlin-android/) · [compose-multiplatform](skills/compose-multiplatform/) · [expo](skills/expo/) · [tauri](skills/tauri/) · [electron](skills/electron/) · [rust](skills/rust/) · [wordpress](skills/wordpress/) · [shopify](skills/shopify/) · [no-code-app](skills/no-code-app/) · [chrome-extension](skills/chrome-extension/) · [api-design](skills/api-design/)
  445. ### 🎮 Game development
  446. [godot](skills/godot/) · [unity](skills/unity/) · [unreal](skills/unreal/) · [game-design](skills/game-design/) · [game-storytelling](skills/game-storytelling/) · [level-design](skills/level-design/) · [gamedev-shaders](skills/gamedev-shaders/) · [gamedev-multiplayer](skills/gamedev-multiplayer/) · [gamedev-physics](skills/gamedev-physics/) · [gamedev-pathing](skills/gamedev-pathing/) · [gamedev-shipping](skills/gamedev-shipping/)
  447. ### 🗄️ Databases & data layer
  448. [postgresdb](skills/postgresdb/) · [mysql](skills/mysql/) · [mongodb](skills/mongodb/) · [redis](skills/redis/) · [supabase](skills/supabase/) · [neon](skills/neon/) · [planetscale](skills/planetscale/) · [sqlite-turso](skills/sqlite-turso/) · [prisma-orm](skills/prisma-orm/) · [drizzle-orm](skills/drizzle-orm/) · [firebase](skills/firebase/) · [dynamodb](skills/dynamodb/) · [vector-db](skills/vector-db/) · [clickhouse-analytics](skills/clickhouse-analytics/) · [duckdb](skills/duckdb/) · [db-migrations](skills/db-migrations/) · [backups](skills/backups/)
  449. ### ☁️ Ship & operate: platforms
  450. [vercel](skills/vercel/) · [netlify](skills/netlify/) · [cloudflare](skills/cloudflare/) · [railway](skills/railway/) · [render](skills/render/) · [fly-io](skills/fly-io/) · [coolify](skills/coolify/) · [hetzner](skills/hetzner/) · [digitalocean](skills/digitalocean/) · [aws-essentials](skills/aws-essentials/) · [gcp-essentials](skills/gcp-essentials/)
  451. ### 🛠️ Ship & operate: devops
  452. [docker](skills/docker/) · [github-actions](skills/github-actions/) · [git-workflow](skills/git-workflow/) · [domains-dns](skills/domains-dns/) · [monitoring](skills/monitoring/) · [email-deliverability](skills/email-deliverability/) · [scaling](skills/scaling/) · [deployment](skills/deployment/) · [deprecation](skills/deprecation/)
  453. ### 🔒 Ship & operate: quality & security
  454. [code-review](skills/code-review/) · [security-audit](skills/security-audit/) · [testing-py](skills/testing-py/) · [testing-web](skills/testing-web/) · [testing-go](skills/testing-go/) · [e2e-testing](skills/e2e-testing/) · [accessibility](skills/accessibility/) · [performance](skills/performance/) · [error-handling](skills/error-handling/) · [observability](skills/observability/)
  455. ### 🌀 Motion & interface craft
  456. [variant-explorer](skills/variant-explorer/) · [scroll-film](skills/scroll-film/)
  457. ### 🎨 Design & content craft
  458. [design](skills/design/) · [ux](skills/ux/) · [presentations](skills/presentations/) · [course-storytelling](skills/course-storytelling/) · [course-builder](skills/course-builder/) · [technical-writing](skills/technical-writing/) · [translation-l10n](skills/translation-l10n/)
  459. ### 🧠 Knowledge & meta
  460. [knowledge-ops](skills/knowledge-ops/) · [codebase-onboarding](skills/codebase-onboarding/) · [research-ops](skills/research-ops/) · [decision-records](skills/decision-records/) · [continuous-learning](skills/continuous-learning/) · [skill-scout](skills/skill-scout/) · [context-budget](skills/context-budget/) · [roast-me](skills/roast-me/) · [fable-operator](skills/fable-operator/)
  461. ---
  462. ## Why you can trust it
  463. The harness is built on ten principles. Three of them shape everything else:
  464. - **Evidence over claims.** A task is done when its check was observed, not when someone believes it
  465. works. The same rule applies to this README: every number here comes from `manifest.json` or the
  466. code, and tests check that the README and the code agree.
  467. - **A gate that checks nothing is worse than no gate.** Every rule rsc declares binding ships with the
  468. mechanism that enforces it and a test that watches that mechanism fail and pass.
  469. - **Friction proportional to risk.** Silence for the trivial, ceremony only where failure is
  470. expensive. A harness that nags on harmless work gets switched off.
  471. **Known limits, stated openly:** the hard guards run only in Claude Code; OpenCode 2 and Windows have
  472. had less real-world testing than macOS and Linux; and the session memory is local by design, never
  473. cloud.
  474. ---
  475. ## Contributing
  476. `skills/<name>/` is the single source of truth. Each skill is a directory whose `SKILL.md` frontmatter
  477. drives triggering and recommendations:
  478. ```yaml
  479. ---
  480. name: my-skill
  481. description: Use when [specific triggers]… NOT x (that is sibling).
  482. tags: [keyword, keyword] # what the consult advisor searches over
  483. recommends: [sibling-skill] # what the system offers to install next
  484. origin: risco
  485. ---
  486. ```
  487. After editing any skill:
  488. ```bash
  489. npm run manifest # regenerate manifest.json from skills/*/SKILL.md
  490. npm run validate # validate frontmatter and recommends integrity
  491. npm test # unit and integration tests
  492. bash scripts/eval-lint.sh # validate every skills/*/evals/cases.yaml
  493. ```
  494. `manifest.json` is generated, never edited by hand; CI fails if it is stale. The rubric every skill is
  495. held to is `scripts/skill-rubric.md`. Bug reports are welcome as GitHub issues. Release notes are in
  496. [GitHub Releases](https://github.com/ericrisco/rsc-harness/releases).
  497. ### Third-party skills
  498. Most of this catalog is written here. These are not, and they keep their author's credit:
  499. | Skills | Source | License |
  500. | --- | --- | --- |
  501. | `design-eng` · `animate` · `animate-expo` · `review-animations` · `improve-animations` · `find-animation-opportunities` · `animation-vocabulary` · `apple-design` · `prototype` · `pick-ui-library` · `write-swift` · `ask-sonner` | [emilkowalski/skills](https://github.com/emilkowalski/skills) by Emil Kowalski, commit `d23d7f8` | MIT |
  502. | the engine of `design` (`skills/design/engine/`: direction draw, browser inspector, study library), shipped unmodified | [Leonxlnx/taste-skill](https://github.com/Leonxlnx/taste-skill) v2.1.4 | MIT |
  503. | the four modes, the fresh review and the `DESIGN.md` record in `design` (text rewritten) | [pbakaus/impeccable](https://github.com/pbakaus/impeccable) by Paul Bakaus, `skill-v4.7.0` | Apache-2.0 |
  504. | the answer-first habits in `orient` | [ayghri/i-have-adhd](https://github.com/ayghri/i-have-adhd) | MIT |
  505. | the scroll-scrubbed camera flight, the frame handoff between clips and the phone fixes behind `scroll-film` (prose and engine rewritten) | [oso95/scroll-world](https://github.com/oso95/scroll-world) | MIT |
  506. Adapted, not mirrored: each carries rsc frontmatter, routing evals and hand-offs to its siblings. The
  507. craft bar in them is the original author's.
  508. ### Shared risk policy and external branch checks
  509. Commit `.rsc-risk.json` at the workspace root to share risk rules across clones and submodules:
  510. ```json
  511. {
  512. "version": 1,
  513. "trunks": ["develop", "integration"],
  514. "repositories": { "Catman": { "trunks": ["develop"] } },
  515. "testRepos": ["CatmanTest"],
  516. "rules": [
  517. { "class": "schema", "tier": 2, "path": "\\.Table\\.al$",
  518. "removed": "^\\s*field\\s*\\(", "productionOnly": true,
  519. "because": "Removing a field changes the stored schema." },
  520. { "class": "api", "tier": 2, "path": "\\.Page\\.al$",
  521. "contains": "PageType\\s*=\\s*API", "removed": "APIVersion",
  522. "because": "The API contract changes." },
  523. { "class": "platform", "tier": 2, "path": "(^|/)app\\.json$",
  524. "jsonKeys": ["dependencies", "idRanges", "runtime"],
  525. "because": "Dependencies or platform requirements change." },
  526. { "class": "editor", "tier": 0, "path": "^\\.vscode/settings\\.json$",
  527. "because": "Editor preferences only." }
  528. ]
  529. }
  530. ```
  531. Rules can also use `added`, `deleted: true` and `repos: ["relative/path"]`. Conditions in one rule
  532. must all match. `contains` reads the file before and after the change; `added` and `removed` read
  533. changed lines with whitespace differences ignored. JSON keys are top-level keys. Renames are
  534. checked as both old and new paths. Matching rules take their highest tier; explicit zero replaces
  535. ordinary-code defaults, but built-in risk classes and the secrets/harness floors still apply.
  536. Patterns are case-insensitive and limited to 200 characters; repeated groups, lookarounds and
  537. backreferences are rejected. Content lines over 2,000 characters are skipped. Rule evaluation has
  538. a hard one-second timeout and an 8 MiB input budget: exceeding either reports an error at tier 2.
  539. At most 64 rules and 16 trunk candidates are supported.
  540. `rsc risk` and the push gate use the policy **from the workspace trunk**, including repository
  541. settings and exemptions. A proposed policy cannot relax its own outgoing branch. A policy first
  542. added on a feature branch stays inactive until it lands. Use `rsc risk --policy-preview` to inspect
  543. the proposed policy explicitly; the output identifies its source. Local `.rsc/sello-config.json`
  544. remains available for existing settings and bootstrap. Initial trunk discovery uses `origin/HEAD`,
  545. then `main`, `master` or `develop`; for another name, set the remote HEAD or the local `trunk` first.
  546. Once a trusted policy exists, its trunk candidates select the nearest branch by commits ahead.
  547. `rsc doctor` reports an installed ASDD gate that cannot find a trunk.
  548. Specs accept `branches: [feature/a, fix/b]` or a YAML block list, as well as the existing `branch:`.
  549. An FTD under `02-DOCS/wiki/ftd/` can link its branch to a spec with `spec: <slug>`. Optional trusted
  550. `asddGate` settings are `require: "spec"` (default: full panel), `exemptBranches: ["^chore/back-merge-"]`
  551. and `cherryPickSuffix: "-master"`. These settings take effect after the policy reaches the trunk.
  552. External PR tools can run `rsc spec gate --branch-check --repo <directory> [--branch <name>]` from
  553. the workspace, or run the same command inside a submodule. Exit `0` means allowed, `2` means denied,
  554. and `1` means the check could not run. `--branch` evaluates that branch even when it is not checked
  555. out. The hook resolves ordinary `git -C`, directory changes, shell `-c`/`eval` wrappers and `gh pr`
  556. commands. Arbitrary scripts, aliases and shell functions need the explicit CLI check; their bodies
  557. cannot be inferred from the shell invocation. This is a local review aid, not a remote-server policy.
  558. Sync preserves opt-outs declared by teammates in `.rsc.json`. Existing local markers remain local
  559. choices; doctor reports differences. To explicitly re-arm a team gate, remove its local marker and
  560. run an explicit install, which records the new decision; sync alone does not erase the declaration.
  561. ## License
  562. MIT. See [LICENSE](LICENSE).
  563. *Made in Andorra*

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.