1---2name: sc4sap-analyze-symptom3description: Read-only, step-by-step root cause analysis for SAP operational errors. Reads the dump summary and failing source via MCP (known-issue web lookup only when the failure is in standard SAP code), widens to transports and where-used only when the clues call for it, narrows hypotheses with minimal user questions, and provides SAP Note search keywords. Never changes code or data.4---56# SC4SAP Analyze Symptom78Performs structured root cause analysis for SAP operational incidents by connecting to the live SAP system through MCP. Auto-collects evidence from dumps, system state, recent transports, and code call graphs before asking the user any question.91011<Purpose>12sc4sap:analyze-symptom is the first-line triage skill for SAP production incidents. Rather than bombarding the user with questions, it **directly investigates the SAP system through MCP** to gather evidence it can collect on its own. It then asks the user only about gaps that MCP cannot fill, narrows hypotheses to 2–3 categories, and produces SAP Note search keywords plus recommended next actions.13</Purpose>1415<Response_Prefix>16Every response triggered by this skill MUST begin with `[Model: <main-model> · Dispatched: <sub-summary>]` per [`../../common/model-routing-rule.md`](../../common/model-routing-rule.md) § Response Prefix Convention.17</Response_Prefix>1819<Phase_Banner>20Multi-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`](../../common/model-routing-rule.md) § Phase Banner Convention.21</Phase_Banner>2223<Use_When>24- User reports a symptom using words like "error", "dump", "failing", "broken", "not working", "timeout", "slow"25- User has at least one clue: error message, TCode, program name, job name, or affected user/data26- User is unsure which log or transaction to inspect (ST22, SM21, SLG1, SU53, SM13, SM58, WE02, etc.)27- Need to classify whether the issue is custom development vs SAP standard28- Need to trace root cause of an incident that started after a recent transport or patch29</Use_When>3031<Do_Not_Use_When>32- Root cause is already identified and only a code fix is needed — use `/sc4sap:create-program` or direct MCP `Update*` calls33- Pure static code quality review — use `/sc4sap:analyze-code`34- Need to create a new ABAP object — use `/sc4sap:create-object`35- Conceptual or configuration-guide question — use a module consultant agent directly36</Do_Not_Use_When>3738<Session_Trust_Bootstrap>39**MANDATORY — runs as Step 0 before any MCP call or user interaction.**4041Invoke `/sc4sap:trust-session` with `parent_skill=sc4sap:analyze-symptom` to pre-grant all MCP tool + file-op permissions for this session (eliminates per-tool "Allow this tool?" prompts during auto-investigation — `RuntimeAnalyzeDump`, `ListTransports`, `GetWhereUsed`, etc.).4243- Headless host (system prompt declares `Host: sc4sap-web`) → skip entirely; the host governs permissions.44- If `.sc4sap/session-trust.log` already has a line within the last 24h, skip silently. Otherwise run it.4546Full spec: see [`../trust-session/SKILL.md`](../trust-session/SKILL.md).47</Session_Trust_Bootstrap>4849<Core_Principles>50- **Read-only, always**: analysis and reading only — no SAP write tools (Create/Update/Delete/Patch/Write/Activate, RunUnitTest, RuntimeRun*/RuntimeCreate*, CreateTransport), no Edit/Write, even when the user asks for the fix. Fixes appear only as proposals.51- **Cheapest evidence first**: dump metadata and termination link, then the one failing include; the ~50 KB formatted dump only when those cannot explain the error. A short web lookup (standard identifiers only — never Z*/Y* names, SID, user IDs, or data) runs only when the failure point is standard SAP code. Never re-ask what MCP can answer.52- **Evidence over assumption**: Do not speculate. No "probably" statements without supporting MCP or user-provided evidence.53- **Minimal questions**: At most 3 questions per round. Skip any question whose answer is already known via MCP.54- **Hypothesis narrowing**: Reduce candidate causes to 2–3 from the 8-category framework; each must carry a confidence level and a confirmation path.55- **Actionable output**: Every hypothesis must include the next evidence step (another MCP call, a TCode, or an escalation target).56- **Customization cache first (local, before live MCP) when a Z*/Y* object or customized SAP include appears in the trace**: read `<CUSTOMIZATION_DIR>/<MODULE>/{enhancements,extensions}.json` (the path Step 1 resolves once — never search for it) and correlate — a `Z*` class in a dump may be a known BAdI impl, a customized `MV45AFZZ`/`ZXRSRU01` may be a recorded form-based exit, a failing field may be a recorded append. Follow `common/customization-lookup.md`. If the cache is absent, suggest `/sc4sap:setup customizations` but do not block the current analysis.57</Core_Principles>5859<Analysis_Framework>60All hypotheses must map to one of these 8 root cause categories:6162| Category | Typical Symptoms | Key Signals |63|----------|------------------|-------------|64| Master / Input data | Only specific data fails, others succeed | Data values, related master records |65| Authorization | Only specific users fail | SU53, STAUTHTRACE, recent role changes |66| Customizing | Only specific org units affected | SPRO values, recent customizing transports |67| Interface / RFC / Batch | External integration fails | SM58, SMQ1/2, SM37, WE02, BD87 |68| Custom development | Z*/Y* objects in call stack | Recent Z* transports, GetWhereUsed |69| Standard SAP bug | Only standard objects in stack; right after SP upgrade | SAP Note search, kernel/SP level |70| Performance / Locks / DB | Timeouts, increased wait times | ST05, SAT, SM12, SQLM |71| Operational procedure | Step order or prerequisite violated | Month-end, dependency job status |7273Every hypothesis presented to the user must declare its **category** explicitly.74</Analysis_Framework>7576<Evidence_Collection_Matrix>77Evidence collection strategy — prefer MCP auto-query, fall back to manual TCode guidance:7879| Symptom Type | MCP Auto-Query | Manual TCode |80|--------------|----------------|--------------|81| Short dump / runtime error — **first** | `RuntimeListFeeds` (`dumps`, time window) when no dump ID is known → `RuntimeGetDumpById` metadata (~7 KB, termination link) → failing include; formatted dump (~50 KB) only when needed | ST22 |82| Known issue — only when the failure point is standard SAP code | `WebSearch` / `WebFetch` (SAP Notes, KBAs, SAP Community) — standard identifiers only | SAP for Me |83| Performance / long runtime | `RuntimeListProfilerTraceFiles`, `RuntimeAnalyzeProfilerTrace` (existing traces only — never start a run) | ST05, SAT, SQLM |84| Suspect program/class logic | `GetInclude`/`GetProgram`/`GetClass`/`GetFunctionModule`, `GetAbapAST`, `GetAbapSemanticAnalysis`, `GetWhereUsed` | SE80, SE24, SE38 |85| Recent change tracking | `ListTransports`, `GetTransport`, `GetObjectInfo` (Author/Changed-by) | SE09, SE10, SE16 → E070 |86| **Z\*/Y\* object or customized SAP include in trace** | Local file read (path from Step 1, no search): `<CUSTOMIZATION_DIR>/<MODULE>/enhancements.json` (→ `badiImplementations[]`, `cmodProjects[]`, `formBasedExits[]`) and `<CUSTOMIZATION_DIR>/<MODULE>/extensions.json` (→ `appendStructures[]`) | n/a — local cache only |87| Enhancement / BAdI | `GetEnhancements`, `GetEnhancementImpl`, `GetEnhancementSpot` | SE18, SE19, SMOD, CMOD |88| System / session info | `GetSession` | /n (status), /o SM04 |89| Table schema (not rows) | `GetTable`, `GetStructure`, `GetView`, `GetDataElement`, `GetDomain` | SE11 |90| Unit test results | `GetUnitTestResult`, `RunUnitTest` | SE80 → test class |91| Authorization error | (MCP not supported) | SU53, STAUTHTRACE |92| Application log | (MCP not supported) | SLG1 |93| System log | (MCP not supported) | SM21 |94| Update error | (MCP not supported) | SM13 |95| RFC / tRFC / qRFC | (MCP not supported) | SM58, SMQ1, SMQ2 |96| Background job | (MCP not supported) | SM37 |97| IDoc | (MCP not supported) | WE02, WE05, BD87 |98| OData / Fiori | (MCP not supported) | /IWFND/ERROR_LOG, /IWBEP/ERROR_LOG |99100**Rule**: For any MCP-supported item, never ask the user — query it directly.101</Evidence_Collection_Matrix>102103<Workflow_Steps>104**MANDATORY**: Follow the step sequence defined in [`workflow-steps.md`](workflow-steps.md).105106Per-step model allocation (skill main thread runs on Sonnet 4.6 per frontmatter; heavy analysis is delegated):107108| Step | Owner | Model | Role |109|------|-------|-------|------|110| 0 Trust | skill-to-skill | Sonnet | permission bootstrap — skipped on a headless host |111| 1 Initial Triage | main | **Sonnet** | clue parsing + `GetSession` + MODE (`quick-dump` \| `full`) + customization path |112| **2 Investigate + Narrow + Report** | **`sap-debugger`** — `model: "sonnet"` for `quick-dump`, `model: "opus"` for `full` | **Sonnet / Opus** | Read-only. `quick-dump`: dump summary → failing source; returns `BLOCKED — needs full` when the dump alone is not enough, and main re-dispatches the round as `full` on Opus. `full`: dump / scoped transports / code / enhancement / customization; web lookup only for standard-code failures. Writes the user-facing report per `output-format.md` (hypotheses, questions, Note keywords, next steps). One dispatch per round. |113| 3 Relay | main | **Sonnet** | output the report verbatim; wait for answers → repeat Step 2 |114| 4 Follow-up Routing | main | **Sonnet** | pointers only when the user asks for the fix (apply the proposal in SE38/ADT outside this skill, /sc4sap:analyze-code, module consultant) — never a write call |115116sap-debugger's tool set already covers the dumps feed (`RuntimeListFeeds`), `RuntimeGetDumpById`, profiler, transport queries, code reads, enhancement lookup, and customization cache reads — see the agent's Investigation_Protocol for the full inventory. A `quick-dump` round is read-mostly work (dump → source → explanation), which `common/model-routing-rule.md` § Tier 1 routes to Sonnet; `full` rounds need cross-file reasoning (dump × transport × source × customization) and stay on Opus.117</Workflow_Steps>118119<Question_Strategy>120**Rule**: max 3 questions per response. Never re-ask what MCP already answered.121122Priority when information is missing:1231. Exact error text + message class/number — the strongest SAP Note search key1242. TCode / App / Program / Job where the error occurs1253. Reproduction conditions (always vs intermittent; user/data/org specificity)126127Situation-specific follow-ups:128- **Authorization suspected**: Does another user succeed with the same input? Any SU53 capture?129- **Batch suspected**: Does manual execution also fail? Any recent variant change?130- **Interface suspected**: Does SM59 Connection Test succeed? What is the IDoc status code (51/52/53/64)?131- **Custom development suspected**: (First run `ListTransports` + `GetWhereUsed`, then) Does TR candidate X match the timing of the incident?132- **Standard bug suspected**: Release/SP auto-detected via `GetSession`. Does the same symptom reproduce on QAS/DEV?133- **Performance suspected**: How much slower than usual? Which resource saturates first — DB / CPU / memory?134</Question_Strategy>135136<Output_Format>137Per-round report template and the final-round consolidated report structure live in [`output-format.md`](output-format.md). Follow it literally for both intermediate rounds and the final analysis.138</Output_Format>139140<MCP_Tools_Used>141Main thread: `GetSession` only (Step 1). Everything else runs inside the `sap-debugger` dispatch, so dump payloads, source and transport lists never sit in the orchestrator context. The authoritative tool list and call limits live in the dispatch prompt in [`workflow-steps.md`](workflow-steps.md) § Step 2: dump (`RuntimeListFeeds` dumps feed, `RuntimeGetDumpById` metadata, formatted view only when needed), one failing include (`GetInclude` / `GetProgram` / `GetClass` / `GetFunctionModule`), scoped transports (`ListTransports`, `GetTransport`, `GetObjectInfo`), `GetWhereUsed`, enhancements, existing profiler traces, DDIC schema reads. Never `GetProgFullCode`, `GetTableContents`, `GetSqlQuery` or any write tool.142</MCP_Tools_Used>143144<Common_Pitfalls_To_Avoid>145- ❌ Asking the user for information MCP can retrieve (system info, program source, recent transports)146- ❌ Firing 4+ questions at once147- ❌ Diagnosing a root cause without an error message in hand148- ❌ Skipping the dump lookup when a dump is suspected and speculating instead (`RuntimeListFeeds` dumps feed — `RuntimeListDumps` can return an empty list)149- ❌ Changing anything — code, data, transports, activation — from this skill, even on request150- ❌ Running the full investigation (transports, where-used) on a plain dump with no change-history signal151- ❌ Searching the filesystem for config/caches, or rewriting the debugger's report in the main thread152- ❌ Deflecting with "contact Basis / dev team" without a concrete checklist and evidence153- ❌ Claiming a standard SAP bug before attempting a SAP Note search154- ❌ Pulling the ~50 KB formatted dump before the metadata and failing include were checked, trusting the `summary` key facts of `RuntimeAnalyzeDump` / `RuntimeGetDumpById` (they pick the wrong chapter), or reading a whole program with `GetProgFullCode`155- ❌ Blaming recent changes without inspecting transport history via `ListTransports`156- ❌ Listing 4+ hypotheses (narrow to 2–3)157</Common_Pitfalls_To_Avoid>158159Task: {{ARGUMENTS}}
babamba2