# Sc4sap:setup

> Plugin setup — detect legacy single-profile state (migrate → multi-profile), create or register a SAP connection profile under ~/.sc4sap/profiles/<alias>/, install abap-mcp-adt-powerup MCP server, optionally install DEV-only ZMCP_ADT_UTILS + ZCL_S4SAP_CM_* ALV OOP handlers (tier-gated), register both PreToolUse hooks (blocklist + tier-readonly-guard), optional SPRO / customizations extraction

- **Type:** Skill
- **Install:** `agentstack add skill-babamba2-superclaude-for-sap-setup`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [babamba2](https://agentstack.voostack.com/s/babamba2)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [babamba2](https://github.com/babamba2)
- **Source:** https://github.com/babamba2/superclaude-for-sap/tree/main/skills/setup

## Install

```sh
agentstack add skill-babamba2-superclaude-for-sap-setup
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# SC4SAP Setup

Use `/sc4sap:setup` as the unified setup and configuration entrypoint for SuperClaude for SAP.

Every response triggered by this skill MUST begin with `[Model:  · Dispatched: ]` per [`../../common/model-routing-rule.md`](../../common/model-routing-rule.md) § Response Prefix Convention.

Multi-phase skill. Before each `Agent(...)` dispatch (including the two conditional error-escalation paths below), emit `▶ phase= () · agent= · model=` per [`../../common/model-routing-rule.md`](../../common/model-routing-rule.md) § Phase Banner Convention.

Setup's happy path runs entirely on the main thread (Haiku 4.5 per frontmatter) — it's configuration work, not code generation. Two step-ranges escalate on failure:

| Source step(s) | Agent | Model | Why |
|----------------|-------|-------|-----|
| **4bis RFC backend** — preflight failure, handler install error, SICF service misconfigured, SM59 test fails, ZRFC ICF handler returns 5xx | **`sap-bc-consultant`** | **Opus 4.7** (frontmatter) | Pure Basis domain — RFC destinations, SICF, SM59, transport of handler class. Basis consultant owns this ground. |
| **5 Reconnect / 6 GetSession / 7 systemInfo persist / 8 GetInactiveObjects** — any failure (HTTP error, auth reject, parse failure, timeout, authorization object missing) | **`general-purpose`** (with `model: "opus"` override) | **Opus 4.7** | 3-layer stack (SAP ADT + MCP server + Claude Code plugin). Most errors are cross-layer (MCP framework bugs, Node runtime issues, profile resolution) — pure SAP consultant has a blind spot. general-purpose + Opus spans `Read`/`Grep`/`Bash`/`WebFetch`/`WebSearch` across all three layers. |

Each escalation emits its own phase banner:
```
▶ phase=4bis.escalate (basis) · agent=sap-bc-consultant · model=Opus 4.7
▶ phase=5-8.escalate (triage) · agent=general-purpose · model=Opus 4.7
```

After the escalation agent returns a diagnosis + remediation checklist, the main thread (Haiku) surfaces it to the user and asks whether to retry the failed step or abort setup.

## Usage

```bash
/sc4sap:setup                  # full setup wizard
/sc4sap:setup doctor           # diagnose installation and SAP connection
/sc4sap:setup mcp              # configure abap-mcp-adt-powerup MCP server
/sc4sap:setup spro             # auto-generate SPRO config from S/4HANA system
/sc4sap:setup customizations   # extract customer Z*/Y* enhancements + extensions
```

## Routing

Process the request by the **first argument only**:

- No argument, `wizard`, or `--force` → run the full setup wizard (Steps 0–13). **Read `wizard-steps.md`** (in this skill folder) and execute the steps defined there in order. Step 0 performs legacy detection + profile bootstrap before any question.
- `doctor` → route to `/sc4sap:sap-doctor` with remaining args
- `mcp` → route to `/sc4sap:mcp-setup` with remaining args
- `spro` → **Active-profile precondition** (decision §4.5): if `/.sc4sap/active-profile.txt` is absent AND `~/.sc4sap/profiles/` is empty, error out with `"No active SAP profile — run /sc4sap:setup (full wizard) first"`. If the pointer is absent but profiles exist, error with `"No active profile — use /sc4sap:sap-option switch "`. Otherwise, run SPRO config auto-generation — **read `spro-auto-generation.md`** and follow its 3 steps.
- `customizations` (also accepts `cust` / `enhancements`) → same active-profile precondition as `spro`. Then run customer enhancement + extension extraction — **read `customization-auto-generation.md`** and follow its 3 steps.

**MANDATORY — runs as Step 0 before any MCP call, file write, or user question.**

Invoke `/sc4sap:trust-session` with `parent_skill=sc4sap:setup` to pre-grant MCP tool + file-op permissions for the session. Setup itself creates ABAP objects on DEV tier profiles (`ZMCP_ADT_UTILS`, `ZCL_S4SAP_CM_*` ALV handlers) and writes multiple local files (`~/.sc4sap/profiles//{sap.env, config.json}`, `/.sc4sap/active-profile.txt`, `.claude/settings.local.json`, hook files) — each of which would otherwise trigger a permission prompt. Note: `trust-session` only allows writes under `.sc4sap/**`; writes under `~/.sc4sap/profiles/**` are routed through `sap-profile-cli.mjs` (a Bash invocation) so they go through the normal permission flow — trust-session cannot blanket-approve user-home paths.

- If `.sc4sap/session-trust.log` already has a line within the last 24h, skip silently.
- Otherwise run it and surface the one-line confirmation.
- `GetTableContents` / `GetSqlQuery` remain prompt-gated even after trust-session — setup does not need them.
- All subsequent `Agent` dispatches within this skill MUST pass `mode: "dontAsk"`.

Full spec: see [`../trust-session/SKILL.md`](../trust-session/SKILL.md).

## Interaction Style (MANDATORY)

> **Ask one question at a time. Never batch questions.**
>
> - Walk through the steps **sequentially** and collect each input **one by one**.
> - Do not advance to the next question until the user has answered the current one.
> - As soon as the user answers, record/apply the value and move to the next step.
> - Even when a single step has multiple fields (e.g., Step 4 SAP connection info), ask **one field per message**.
>   Example: ask `SAP_URL?` → wait for answer → ask `SAP_CLIENT?` → wait → ask `SAP_AUTH_TYPE?` → ...
> - If a value already exists in `.sc4sap/sap.env` or `.sc4sap/config.json`, show the current value and offer "press Enter to keep / type a new value".
> - Keep prompts short. Do not paste long wall-of-text instructions.
>
> Never dump the entire questionnaire in a single message — users find it frustrating.

## Companion files (MUST read when relevant)

- `wizard-steps.md` — the 14-step setup wizard **index** (Step 0–13). Read before running the full wizard. Heavy steps are split out (see below).
- `wizard-step-02-system-identification.md` — Step 2 detail (SAP version, ABAP release, Industry).
- `wizard-step-04-profile-creation.md` — Step 4 detail (alias + `SAP_TIER` + connection fields → `sap-profile-cli.mjs add`). Read before Step 4.
- `wizard-step-09-abap-objects.md` — Step 9 detail, including the **DEV-only tier gate** (QA/PRD REFUSE → CTS import guidance).
- `wizard-step-11-optional-extraction.md` — Step 11 (SPRO prompt) + 11b (Customization prompt). Artifacts land under `work//`.
- `wizard-step-12-blocklist-hook.md` — Step 12 installs BOTH L1 PreToolUse hooks: `block-forbidden-tables.mjs` + `tier-readonly-guard.mjs`.
- `spro-auto-generation.md` — SPRO extraction workflow. Read when the first arg is `spro` or when Step 11 proceeds.
- `customization-auto-generation.md` — customer Z*/Y* enhancement + extension extraction. Read when the first arg is `customizations` (also `cust` / `enhancements`) or when running wizard step 11b.
- `rfc-backend-selection.md` — Step 4bis RFC backend selection (soap/native/gateway/odata/zrfc). Writes `SAP_RFC_*` to the ACTIVE PROFILE env, never to the project folder.
- `odata-classes-install.md` — Step 9c OData backend classes install (DEV tier only).
- `hud-statusline.md` — HUD status line spec. HUD reads `/.sc4sap/active-profile.txt` + profile env to render alias + tier badge.
- **External references**: [`../../docs/multi-profile-design.md`](../../docs/multi-profile-design.md), [`../../docs/multi-profile-setup-gap.md`](../../docs/multi-profile-setup-gap.md), [`../../common/multi-profile-artifact-resolution.md`](../../common/multi-profile-artifact-resolution.md), [`../sap-option/migration.md`](../sap-option/migration.md), [`../sap-option/profile-management.md`](../sap-option/profile-management.md).

## HUD Status Line

The plugin ships a sc4sap-branded status line that activates automatically on install (declared in `.claude-plugin/plugin.json` → `statusLine`). **Read `hud-statusline.md`** for the full specification, including displayed segments, environment variables (`SC4SAP_WEEKLY_LIMIT_USD`, `NO_COLOR`), performance characteristics, and how to disable it.

## Notes

- `/sc4sap:sap-doctor`, `/sc4sap:mcp-setup`, `/sc4sap:sap-option` remain valid direct entrypoints. Prefer `/sc4sap:setup` in documentation and user guidance.
- **Configuration layout (0.6.0 multi-profile)**:
  - **User home** (profile definitions, shared across repos):
    - `~/.sc4sap/profiles//sap.env` — MCP-server env; password is stored in the OS keychain and referenced via `SAP_PASSWORD=keychain:sc4sap//`.
    - `~/.sc4sap/profiles//config.json` — plugin-side settings (`sapVersion`, `abapRelease`, `industry`, `activeModules`, `namingConvention`, `systemInfo`, `activeTransport`, `blocklistProfile`).
    - `~/.sc4sap/profiles//.abap-utils-installed` — Step 9 sentinel (DEV only).
    - `~/.sc4sap/profiles/.trash/-/` — soft-deleted profiles (7-day auto-purge).
  - **Project root** (engagement-scoped artifacts):
    - `/.sc4sap/active-profile.txt` — alias of the currently active profile.
    - `/.sc4sap/work//{program, cbo, customizations, audit, comparisons, spro-config.json}` — per-profile artifacts (see [`../../common/multi-profile-artifact-resolution.md`](../../common/multi-profile-artifact-resolution.md)).
    - `/.sc4sap/sap.env.legacy` — migration backup (rollback); exists only right after Step 0 migration.
    - Legacy `/.sc4sap/{sap.env, config.json}` — DELETED by Step 0 migration (decision §4.3). Fresh installs never create them.
    - Optional `/.sc4sap/blocklist-{extend,custom}.txt` — L1 hook extension lists.
- **Post-setup edits**:
  - Change connection / password / industry / blocklist profile / RFC backend → `/sc4sap:sap-option`.
  - Switch active system → `/sc4sap:sap-option switch `.
  - Add another company or tier → `/sc4sap:sap-option add`.
  - Tier is immutable on a profile — change by remove + add via `/sc4sap:sap-option`.
- **Tier semantics** (enforced by `scripts/hooks/tier-readonly-guard.mjs` + L2 MCP guard):
  - `DEV` — writes allowed; Step 9 installs on this tier only.
  - `QA` — read + `RunUnitTest` only; mutations blocked; Step 9 refuses.
  - `PRD` — strict read-only; `RunUnitTest` blocked; Step 9 refuses.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [babamba2](https://github.com/babamba2)
- **Source:** [babamba2/superclaude-for-sap](https://github.com/babamba2/superclaude-for-sap)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** yes
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-babamba2-superclaude-for-sap-setup
- Seller: https://agentstack.voostack.com/s/babamba2
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.