- name
SKILL.md
스킬 파일은 게시자의 저장소에 그대로 있습니다.
복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add matlab/simulink-agentic-toolkit --skill slci-inspection-incompatibility-analysis -a cursor -y복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add matlab/simulink-agentic-toolkit --skill slci-inspection-incompatibility-analysis -a claude-code -y복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add matlab/simulink-agentic-toolkit --skill slci-inspection-incompatibility-analysis -a codex -y복사한 명령을 터미널에 붙여넣습니다. 이 SKILL.md가 해당 에이전트에 설치됩니다.
npx skills add matlab/simulink-agentic-toolkit --skill slci-inspection-incompatibility-analysis -a grok -y복사한 메시지를 Grok Bot 채팅에 보냅니다. Bot이 GitHub에서 이 SKILL.md를 가져와 모든 Bot이 공유하는 스킬에 추가합니다.
Install this skill: https://github.com/matlab/simulink-agentic-toolkit/tree/HEAD/skills-catalog/code-generation/slci-inspection-incompatibility-analysis게시자 제공
Analyze SLCI code inspection results and trace failures back to compatibility constraint violations or known limitations by matching failed code lines against known incorrect code generation patterns and documented limitation scenarios.
slci-compatibility-analysis insteadAsk the user whether the inspection failure might be related to a compatibility issue before taking any other action. Do not search for files or run MATLAB until the user confirms.
Get the report folder from MATLAB:
c = slci.Configuration('<model_name>');
reportFolder = c.getReportFolder();
Look for <model_name>_report.html (or .pdf/.doc). If found, compare the
model's timestamp against the report's — warn if the report may be stale.
If the report does NOT exist, stop and warn the user. Suggest c.inspect().
If the user explicitly asks to run inspection, first check whether generated code
is stale or missing and regenerate if needed before inspecting.
Code regeneration commands (used throughout this skill):
rtwbuild('<model_name>')c.setTopModel(false); slbuild('<model_name>', 'ModelReferenceRTWTargetOnly')Pre-inspection settings: If the user asks for advanced loop analysis support,
set c.setApplyAdvancedLoopSupport(true) before calling c.inspect(). This
enables verification of algebraic loop constructs that SLCI skips by default.
Extract the text content and search for items with Failed, Error, or Unverified status, capturing 2 lines of context before and 5 lines after each match.
For each failed item, identify:
For each failed code location, read the generated C source file at the failed line, capturing 5–10 lines of surrounding context to understand the pattern.
Generated code is located at:
<model_dir>/<model_name>_ert_rtw/<source_file>.c<model_dir>/slprj/ert/<model_name>/<source_file>.cFor Stateflow failures, consult stateflow_failure_patterns.md for traceability comment format, merge junction patterns, and a quick-triage table before reading full constraint reference files.
Read the constraint files from the slci-compatibility-analysis skill's
references/ folder.
See the Constraint Categories table in slci-compatibility-analysis for the
full list of constraint files and their categories.
For each failed code line, use the block path from its traceability comment to identify the block type, then compare the code against the constraint descriptions and Why explanations in the matching constraint files.
Critical: Verify applicability before declaring a match. A structural resemblance alone is NOT sufficient. For each candidate match, verify:
If no constraint satisfies all three checks, report the unmatched failures to the user and ask whether to proceed to Step 4b (limitation matching). If the user declines, classify those failures as UNMATCHED and include them in the report with the constraints that were considered and why each was ruled out. Do NOT check limitations without explicit user approval.
Only run this step after the user has approved proceeding from Step 4.
Compare unmatched failures against known limitations in
the slci-compatibility-analysis skill's references/code_inspection_limitations.md.
Constraint violations take priority — only classify as LIMITATION if no
constraint is a better explanation. If the failure matches neither a constraint
nor a limitation, classify as UNMAPPED and document why each candidate was
ruled out. Do NOT report a "closest match" as the root cause — an incorrect
mapping is worse than an honest "unmapped."
Skip this step if inspection was terminated (e.g., missing headers or compile errors).
Use slci-compatibility-analysis to obtain or verify compatibility results.
Cross-reference the compatibility warnings against the matched constraints to
confirm the mapping. When a warning could map to multiple conditions, check the
model's actual settings to identify the true trigger rather than relying on the
warning text alone.
In the context window, show only a brief summary table (columns: #, Location, Pattern, Root Cause, Type) and summary counts.
Save detailed results to:
<model_dir>/<model_name>_inspection_analysis_YYYY-MM-DD.md
If constraint violations were found, ask the user if they want to apply fixes.
If yes, delegate to the slci-compatibility-analysis skill to apply fixes, then
regenerate code and re-run c.inspect(). Do not fix the model yourself — no
set_param, no fixIncompatibilities, no model modifications of any kind.
If failures remain after fixes, re-run this analysis on the fixed model. If only UNMATCHED, UNMAPPED, or LIMITATION failures remain, inform the user that automatic fixing is not possible.
If no failed items are found, inform the user that all checks passed and save a brief report. If compatibility warnings exist despite passing, include a Warning Impact Analysis explaining why each warning did not cause a failure and under what conditions it could become one.
slci-compatibility-analysis. If that skill is unavailable, stop.Copyright 2026 The MathWorks, Inc.
>
---name: slci-inspection-incompatibility-analysisdescription: >---# skills-catalog/code-generation/slci-inspection-incompatibility-analysis/SKILL.mdFAQ
게시자에게서 skills-catalog/code-generation/slci-inspection-incompatibility-analysis/SKILL.md를 다운로드하거나 경로를 복사해 저장소를 엽니다. PlugMCP는 스킬을 목록에 올릴 뿐 파일을 호스팅하지 않습니다.
게시자가 https://github.com/matlab/simulink-agentic-toolkit에서 관리합니다.
Claude, Cursor, Codex, Grok Build는 게시자의 저장소에서 추가한 SKILL.md를 읽을 수 있습니다. Grok Bot은 채팅에서 스킬의 GitHub 폴더 링크를 보내면 설치합니다.