From moe-execution
Engineering workflow orchestrator. Use `workflow-gate:init` to patch mandatory workflow rules into CLAUDE.md/AGENTS.md/.github/copilot-instructions.md (and AGENT.md if present). Use `workflow-gate:impact-analysis` before changing any public API, interface, shared type, DB schema, exported signature, or symbol with more than 3 call sites. For coding/refactoring/debugging execution, invoke `moe-execution` after planning and impact analysis.
How this skill is triggered — by the user, by Claude, or both
Slash command
/moe-execution:workflow-gateThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
`$ARGUMENTS`:
$ARGUMENTS:
init — bootstrap/patch engineering workflow rules in project instruction filesimpact-analysis — run mandatory blast-radius analysis before interface changesIf $ARGUMENTS is empty:
init when user asks to set up workflow guardrails.impact-analysis when user is about to change interfaces/contracts.workflow-gate as gatekeeper only, then hand off to moe-execution after planning (+ impact-analysis when required).initRun:
python3 .claude/skills/workflow-gate/scripts/init_engineering_workflow.py
Expected behavior:
CLAUDE.md, AGENTS.md, AGENT.md, .github/copilot-instructions.md.moe-execution.After plan + (if needed) impact analysis:
moe-execution for implementation/refactor/debug execution.workflow-gate responsible for pre-change gates and workflow ordering.moe-execution responsible for MoE routing stability and per-change verification discipline.impact-analysisNo interface change without blast-radius report first.
Search these layers in order:
Project hazard checks (hidden dependencies):
| Hazard | Why it hides callers | Check |
|---|---|---|
| jOOQ generated code | Schema changes break generated classes | rg "FROM.*table_name" src/ |
| Flyway migrations | Later migrations depend on earlier table shape | ls -1 src/main/resources/db/migration/ | sort |
| Spring bean wiring | @Qualifier/bean names wire by string | rg "Qualifier.*beanName" --type java |
| Temporal activities | Activity names called by string | Search workflow files for activity name strings |
| Kafka schemas | Producer/consumer drift | Check producer and consumer schema usage |
| TypeScript API mirrors | Frontend DTOs mirror backend manually | rg "interface.*SameName" frontend/src/ |
| MCAP/ROS message types | Bag/clip format tied to ROS definitions | Check bagdb-mcap/ and clip format docs |
| OpenSearch mappings | Mapping changes break queries/dashboards | Check OpenSearch query templates |
Classify dependents:
Rate risk: LOW / MEDIUM / HIGH / CRITICAL.
Report before code edits:
Blast radius: <symbol>
d=1 WILL BREAK: <N> callers
d=2 LIKELY AFFECTED: <N> consumers
d=3 TEST: <N> suites/paths
Risk: <LOW|MEDIUM|HIGH|CRITICAL>
BC path: <Yes/No + strategy>
Plan: <ordered steps>
Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
npx claudepluginhub uuie/deepseek-instructions --plugin moe-execution