sc4sap:analyze-cbo-obj

SC4SAP Analyze CBO Objects

Walks a CBO (Customer Business Object) package, inventories every project-built ABAP element (table, structure, data element, class, interface, function module, program, view, table type), detects which elements are frequently reused inside the package, infers each element's business purpose from its name/fields/descriptions, and persists the result to .sc4sap/cbo/<MODULE>/<PACKAGE>/ for downstream skills (program, program-to-spec, create-object, autopilot) to consult before creating anything new.

<Response_Prefix> Every response triggered by this skill MUST begin with [Model: <main-model> · Dispatched: <sub-summary>] per ../../common/model-routing-rule.md § Response Prefix Convention. </Response_Prefix>

<Phase_Banner> Multi-phase skill. Before each Agent(...) dispatch, emit ▶ phase=<id> (<label>) · agent=<name> · model=<Opus 4.7|Sonnet 4.6|Haiku 4.5> per ../../common/model-routing-rule.md § Phase Banner Convention. </Phase_Banner>

<Use_When>

• Starting development on a module that already has a sizeable Z-package
• Onboarding onto an AMS / support engagement (need a map of custom assets)
• Before /sc4sap:create-program or /sc4sap:create-object on a new spec — so reuse is evaluated
• User says "analyze CBO", "analyze custom objects", "map Z package", "list frequently used customs", "CBO inventory" </Use_When>

<Do_Not_Use_When>

• User wants a code quality review of one object → /sc4sap:analyze-code
• User wants to reverse-engineer ONE program into a spec → /sc4sap:program-to-spec
• User wants to create an object → /sc4sap:create-object
• Package does not yet contain custom objects (CBO discovery is not meaningful) </Do_Not_Use_When>

<Session_Trust_Bootstrap> MANDATORY — runs as Step 0 before any MCP call or user interaction.

Invoke /sc4sap:trust-session with parent_skill=sc4sap:analyze-cbo-obj to pre-grant all MCP tool + file-op permissions for this session (eliminates per-tool "Allow this tool?" prompts during the 8-step walk).

• If .sc4sap/session-trust.log already has a line within the last 24h, skip silently.
• Otherwise run it and surface the one-line confirmation.

Full spec: see ../trust-session/SKILL.md. </Session_Trust_Bootstrap>

<Workflow_Steps> Orchestration is 3 main-thread Socratic steps (Haiku) + one delegated dispatch to sap-stocker (Sonnet) + a branching hand-off. Detailed spec lives in workflow-steps.md.

• Step 1 / 1.5 / 2 (main thread · Haiku) — Socratic intake: package name → flagship programs (optional <KEY_PROGRAMS>) → module. Frontmatter model: haiku pins the main thread to Haiku 4.5 for cost-efficient Q&A — no domain judgment required for intake.
• Step 3–7 (delegated · Sonnet 4.6) — One Agent(...) dispatch to sap-stocker. The stocker runs its own Investigation_Protocol: walk → GetWhereUsed graph → pin/frequency tiering → business-purpose inference → cross-module gap analysis (per ../../common/active-modules.md) → sensitive-name flagging → persist index.md + inventory.json → return a Logic-heavy: <bool> flag. Authoritative spec: ../../agents/sap-stocker.md § Investigation_Protocol + § Output_Format.
• Step 8 (branching):
  • Branch A (Logic-heavy: false, DDIC-dominant) — canned summary printed by main thread (Haiku). No agent dispatch.
  • Branch B (Logic-heavy: true, FM/CLAS/INTF/large-PROG in inventory) — dispatch sap-writer (Haiku 4.5) for a reader-facing briefing: pinned highlights · business-logic assets · cross-module gaps · sensitive objects · next-step hint. Writer BLOCKED → fallback to Branch A.