agentleFS
rahmanef63rahmanef63/si-coder-agent/skills/sc-install/SKILL.md

sc-install

Skill13 starsChanged 2 months agoMITView on GitHub

Install or share SI-Coder across Claude Code, Claude.ai, ChatGPT, Codex, Hermes, OpenClaw, or another Agent Skills-compatible runtime. Use the canonical skill directory containing SKILL.md for local/repository installs. Use ZIP for web upload surfaces that document ZIP. Treat .skill as optional only when the target client explicitly supports that extension. Choose the install path by surface and never promise an invocation syntax the product does not support.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep durable instructions in English.

    line 16
  2. Distinctive:

    Do not give every install method.

    line 20
  3. Distinctive:

    If the user wants only one skill, Claude's skill installer can install a GitHub skill path instead.

    line 33
  4. Distinctive:

    Use the generated package:

    line 37
  5. Very specific:

    Do not require a VPS or local SC vault for the hosted path.

    line 44
  6. Very specific:

    Use the built-in $skill-installer and the GitHub repository paths listed in AI_INSTALL.md.

    line 48
  7. Very specific:

    Use --with-mcp when the user also wants the local SI-Coder MCP server.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

sc-install — install SI-Coder on the current AI surface

Language

Keep durable instructions in English. Reply in the user's language unless they request another language.

First rule: detect the surface

Do not give every install method. Choose the current surface and give the shortest supported path.

Claude Code

Preferred full repository install:

/plugin marketplace add rahmanef63/si-coder-agent
/plugin install si-coder@si-coder-marketplace

The repository provides .claude-plugin/marketplace.json and .claude-plugin/plugin.json. The full plugin route installs the SI-Coder skill set and can load the bundled MCP configuration. After install, the main direct invocation is /sc.

If the user wants only one skill, Claude's skill installer can install a GitHub skill path instead. Do not replace the full plugin install with a single skill when the user wants the whole SI-Coder system.

Claude.ai / Claude Web

Use the generated package:

  • dist/sc.zip — recommended Claude Web upload package (Anthropic documents ZIP upload)
  • dist/sc.skill — optional ZIP-format archive only for clients that explicitly accept the .skill extension

Claude's web Skills UI accepts uploaded custom skills. The current documented flow is Customize → Skills → + → Create skill → Upload a skill. The package is self-contained and bundles the core SI-Coder workflows.

Do not require a VPS or local SC vault for the hosted path. Do not promise /sc in Claude web unless the current UI actually exposes slash invocation; Claude can automatically activate installed skills.

Codex CLI / Codex app

Use the built-in $skill-installer and the GitHub repository paths listed in AI_INSTALL.md. Core paths:

  • skills/sc
  • skills/sc-build
  • skills/sc-all
  • skills/sc-provider
  • skills/sc-install

Codex skill invocation is surface-specific. Use the current Codex skill picker/invocation UX; do not rewrite it as Claude's /sc.

Local fallback requires Node.js 22, 24, or 26 and installs only active/default skills from skills/catalog.json:

bash install.sh --agent codex

Use --with-mcp when the user also wants the local SI-Coder MCP server.

ChatGPT Web

Use uploaded ChatGPT Skills only when the account/workspace exposes the feature. OpenAI currently documents Skills for eligible Business, Enterprise, Healthcare, and Edu workspaces, subject to workspace settings and product availability. When available, use the complete dist/sc.zip package by default. OpenAI documents skill upload and the SKILL.md standard but does not currently require the .skill extension. Use dist/sc.skill only if that ChatGPT surface explicitly accepts or requests it.

ChatGPT currently documents two invocation modes:

  1. automatic use when the skill is relevant, and
  2. explicit selection by @-mention / the skill picker.

For an uploaded Skill, the OpenAI display name is sc, so say @sc (or select the installed Skill) when explicit invocation is needed. For the workspace marketplace plugin, the plugin display name remains SI-Coder, so explicit plugin selection may appear as @SI-Coder. Do not promise /sc on ChatGPT Web unless OpenAI adds and documents that surface behavior.

ChatGPT also has a Plugin Directory. Plugins can bundle Skills and Apps. Publishing SI-Coder as a public ChatGPT plugin is a separate review/distribution lifecycle from uploading a Skill package.

Hermes / OpenClaw / generic local Agent Skills

Local installer requirement: Node.js 22, 24, or 26. It installs active/default skills only; stub and legacy skills are not exposed as normal capabilities.

Skills only:

bash install.sh --agent hermes
bash install.sh --agent openclaw
bash install.sh --skills-dir /path/to/skills

Skills + SI-Coder tool calling:

bash install.sh --agent hermes --with-mcp
bash install.sh --agent openclaw --with-mcp

The installer registers the same scripts/sc-mcp.js server used by the machine/functions.json tool SSOT. The adapter supports MCP 2026-07-28 discovery/stateless calls and preserves 2025-11-25 initialize compatibility. Hermes uses hermes mcp add; OpenClaw uses openclaw mcp add. The MCP surface is user/connection-scoped (sc.user.*) and never accepts plaintext credential values. Use sc.user.connection.request/manage/list for labeled account/auth selection and sc.user.credential.request for direct credential create/rotate handoff; values are entered only through hidden local terminal input or another explicitly secure credential channel.

Package contract

The editable source remains each skills/<name>/ directory with its required SKILL.md. .zip and .skill are generated transport artifacts, not a second source of truth. Rebuild it with:

npm run package:skills

Never hand-edit dist/sc.skill.

Security

Installing SI-Coder must not copy provider secrets. Fresh local onboarding uses sc setup → user → provider → named connection; direct values go only to the selected 0600 connection file. Legacy profile/~/.bashrc helpers are migration-only. Credential/account authorization otherwise happens through secure connected accounts or the local hidden-secret flow.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.