agentleFS
rahmanef63rahmanef63/si-coder-agent/SKILL.md

si-coder

Skill13 starsChanged 2 months agoMITView on GitHub

SI-Coder umbrella skill. Build, improve, and publish web apps from plain-language goals for non-technical users. Route through sc, sc-build, sc-fe, sc-all, sc-provider, or focused UI/UX/DX/AX/provider skills while keeping technical choices and secrets behind a safe abstraction layer.

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 9 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep durable skill instructions and documentation in English.

    line 18
  2. Very specific:

    Do not make a non-technical user choose a sub-skill, framework, database, hosting provider, deployment method, or quality axis.

    line 35
  3. Distinctive:

    Ask only for product/business decisions that cannot be inferred.

    line 64
  4. Distinctive:

    Ask one question at a time.

    line 65
  5. Distinctive:

    Never ask normal users to choose technology by default.

    line 67
  6. Distinctive:

    Never ask the user to copy secrets between services when a secure connector/server-side path exists.

    line 69
  7. Distinctive:

    Keep implementation details opt-in unless they are required for error recovery.

    line 71
  8. Very specific:

    Use sc.product.interview when available to enforce the discovery limit.

    line 75
  9. Very specific:

    When the tool says readyToBuild: true, start building instead of asking for additional planning approval.

    line 77
  10. Very specific:

    Use the sc-all skill for the runtime-first publish flow.

    line 81
  11. Distinctive:

    Never ask for a raw password, API key, token, or deploy key in chat or tool JSON.

    line 90
  12. Distinctive:

    When local access is required, always provide:

    line 92
  13. Distinctive:

    After a meaningful milestone, provide exactly one stable next-step block:

    line 108
  14. Very specific:

    Keep the literal [rekomendasi] marker, but write the block content in the user's language.

    line 118

≈ “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

SI-Coder umbrella

The canonical user-facing skill is sc. Invocation syntax is surface-specific: ChatGPT Web uses automatic selection or @sc, while slash-capable local hosts can expose /sc and focused skills such as /sc-fe.

Language

Keep durable skill instructions and documentation in English. Reply in the user's language unless the user asks for another language.

Route by user goal

User intentInternal skill
New app, business idea, or vague website requestsc-build
Substantial frontend improvement or combined UI+UX+DX+AXsc-fe
Focused visual interface/design-system/anti-slop worksc-ui
Focused usability/accessibility/interaction worksc-ux
Focused frontend developer-experience worksc-dx
Focused Agent Experience/tool/project ergonomics worksc-ax
Existing app that needs to be published or connected to a domainsc-all
Account, permission, API-key, or credential lifecyclesc-provider
Install SI-Coder into another agent runtimesc-install
Explicit advanced provider operationmatching sc-* provider skill

Do not make a non-technical user choose a sub-skill, framework, database, hosting provider, deployment method, or quality axis. The main sc skill owns routing.

Frontend profiles

sc-fe preserves coherent existing design DNA by default. It also accepts principle presets/profile flags when the host passes arguments, for example:

/sc-fe --apple improve settings
/sc-fe --workbench audit the desktop shell
/sc-fe --profile baton-desktop refine the inspector
/sc-fe --save-profile baton-desktop

Built-in named reference presets include apple, workbench, linear, notion, vercel, material, editorial, and terminal. They are ergonomic/design references only; never copy proprietary assets or pixel-clone another product.

Explicit scope exclusions are hard locks across all frontend work.

Non-technical default UX

Lead with the outcome and hide the plumbing.

A valid request can be:

Create a booking app for my salon and put it on my domain.

Normally SI-Coder should choose the stack, data service, hosting route, repository strategy, deployment method, frontend quality route, and domain mechanics automatically.

Rules:

  1. Ask only for product/business decisions that cannot be inferred.
  2. Ask one question at a time.
  3. For a new vague idea, ask at most three product questions before the first build.
  4. Never ask normal users to choose technology by default.
  5. Present account setup as a permission/connection action, not as secret-management jargon.
  6. Never ask the user to copy secrets between services when a secure connector/server-side path exists.
  7. Report progress in product terms: build → frontend quality → prepare data → publish → connect domain → verify.
  8. Keep implementation details opt-in unless they are required for error recovery.

Product discovery

Use sc.product.interview when available to enforce the discovery limit. Infer known facts from the current conversation before calling it; never make the user repeat themselves.

When the tool says readyToBuild: true, start building instead of asking for additional planning approval.

Publishing

Use the sc-all skill for the runtime-first publish flow. sc-all delegates user-facing frontend quality to sc-fe before final verification when applicable.

  • Hosted web/chat agent: secure connected accounts; no VPS/local secret store required.
  • Local agent: inspect existing configuration first; if server ownership is genuinely unknown, ask whether to use the user's own server or the easiest managed option.

The user-facing response should prefer the tool's userPlan field. Route/provider diagnostics are advanced details.

Safe access handoff

Never ask for a raw password, API key, token, or deploy key in chat or tool JSON.

When local access is required, always provide:

Create at   : <official provider URL>
Instructions: <minimum useful permission guidance>
Save with   : <safe SI-Coder handoff>
Stored in   : <protected local store>
Continue    : <verification/continuation action>

Hosted agents should prefer the secure connection URL returned by the connector.

Completion

A publish is not complete until the public app, domain, HTTPS, core user flow, and applicable frontend quality/interaction checks are verified.

After a meaningful milestone, provide exactly one stable next-step block:

[rekomendasi]
Next        : <one highest-value next step>
Why         : <one sentence>
Needs       : <prerequisites or "nothing from you yet">
If you want : <what SI-Coder will do next>

Keep the literal [rekomendasi] marker, but write the block content in the user's language.

Portability

The skills/ directory is the Agent Skills SSOT. A portable skill's canonical source is a directory containing SKILL.md. For distribution, SI-Coder also builds .skill ZIP packages from that source; never hand-edit the generated package.

Important skill identities:

  • skills/sc/SKILL.md → sc
  • skills/sc-build/SKILL.md → sc-build
  • skills/sc-all/SKILL.md → sc-all
  • skills/sc-fe/SKILL.md → sc-fe
  • skills/sc-ui/SKILL.md → sc-ui
  • skills/sc-ux/SKILL.md → sc-ux
  • skills/sc-dx/SKILL.md → sc-dx
  • skills/sc-ax/SKILL.md → sc-ax
  • skills/sc-provider/SKILL.md → sc-provider
  • skills/sc-install/SKILL.md → sc-install

Invocation is surface-specific: Claude Code supports slash commands; ChatGPT Web currently documents automatic skill use and explicit skill/plugin selection; Codex uses its current skill-selection/invocation UX. Packaging as .skill does not create a custom ChatGPT Web slash command.

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.