菜单
Intelligent agent routing -- automatically matches tasks to the best specialist agent based on file patterns, intent keywords, and domain context. Loaded every session to give Claude explicit routing rules for all 58 AgentSpec agents.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Phase 1 of the SDD workflow: capture and validate requirements in one pass. Owns the full Define methodology — classify the input (BRAINSTORM document, meeting notes, email thread, conversation, or direct request), extract entities (problem, users, MoSCoW goals, success criteria, acceptance tests, constraints, out of scope, assumptions), gather technical and data engineering context, calculate the clarity score against the 12/15 gate, fill gaps with targeted questions, and generate the DEFINE document from the shared template with correct status transitions. Use when the user wants to capture requirements, define the feature, structure project scope, or compute a clarity score — "capture requirements", "define the feature", "clarity score", "Phase 1", /define. Do not use for open-ended exploration of ideas and approaches (that is sdd-brainstorm, Phase 0) or for architecture and technical specification (that is sdd-design, Phase 2).
Create the architecture and technical specification for a defined feature — Phase 2 of the SDD workflow. Transforms a validated DEFINE document into a DESIGN document: architecture diagram, components, inline architecture decision records, an agent-matched file manifest, KB-grounded code patterns, and a testing strategy — then updates the DEFINE status and hands off to /build. Use when requirements are captured and technical design is needed — "design the architecture", "technical design", "Phase 2", or /design pointed at a DEFINE_*.md file. Not for capturing or clarifying requirements (that is sdd-define, Phase 1) and not for implementing code (that is sdd-build, Phase 3).
Update an existing SDD phase document (BRAINSTORM/DEFINE/DESIGN) mid-stream with full cascade awareness: classify the change (additive, modifying, removing, architectural), apply it with a version bump and change note, walk the cascade rules for downstream impact, and confirm with the user before applying any cascading edits. Owns the SDD change-management methodology: cascade rules, version-tracking and status-field obligations, and the iterate-vs-restart thresholds. Use when the user wants to update the requirements or design mid-stream — "requirements changed", "the design needs to change", "cascade the change" — or any existing phase document must change after downstream work has started. Do not use for a brand-new feature or a fundamentally different problem: a different problem, different users, or a change beyond roughly half the document calls for sdd-brainstorm/sdd-define instead of an iterate.
Phase 1 of the SDD workflow: capture and validate requirements in one pass. Owns the full Define methodology — classify the input (BRAINSTORM document, meeting notes, email thread, conversation, or direct request), extract entities (problem, users, MoSCoW goals, success criteria, acceptance tests, constraints, out of scope, assumptions), gather technical and data engineering context, calculate the clarity score against the 12/15 gate, fill gaps with targeted questions, and generate the DEFINE document from the shared template with correct status transitions. Use when the user wants to capture requirements, define the feature, structure project scope, or compute a clarity score — "capture requirements", "define the feature", "clarity score", "Phase 1", /define. Do not use for open-ended exploration of ideas and approaches (that is sdd-brainstorm, Phase 0) or for architecture and technical specification (that is sdd-design, Phase 2).
Create the architecture and technical specification for a defined feature — Phase 2 of the SDD workflow. Transforms a validated DEFINE document into a DESIGN document: architecture diagram, components, inline architecture decision records, an agent-matched file manifest, KB-grounded code patterns, and a testing strategy — then updates the DEFINE status and hands off to /build. Use when requirements are captured and technical design is needed — "design the architecture", "technical design", "Phase 2", or /design pointed at a DEFINE_*.md file. Not for capturing or clarifying requirements (that is sdd-define, Phase 1) and not for implementing code (that is sdd-build, Phase 3).
Update an existing SDD phase document (BRAINSTORM/DEFINE/DESIGN) mid-stream with full cascade awareness: classify the change (additive, modifying, removing, architectural), apply it with a version bump and change note, walk the cascade rules for downstream impact, and confirm with the user before applying any cascading edits. Owns the SDD change-management methodology: cascade rules, version-tracking and status-field obligations, and the iterate-vs-restart thresholds. Use when the user wants to update the requirements or design mid-stream — "requirements changed", "the design needs to change", "cascade the change" — or any existing phase document must change after downstream work has started. Do not use for a brand-new feature or a fundamentally different problem: a different problem, different users, or a change beyond roughly half the document calls for sdd-brainstorm/sdd-define instead of an iterate.
| name | agent-router |
| description | Intelligent agent routing -- automatically matches tasks to the best specialist agent based on file patterns, intent keywords, and domain context. Loaded every session to give Claude explicit routing rules for all 58 AgentSpec agents. |
Explicit routing rules for matching tasks to the correct specialist agent. Generated from each agent's frontmatter, so any change to an agent's description, kb_domains, or escalation_rules flows here automatically.
Agent count: 58 | Categories: 8 | Content hash: d2970b1b988f
System-level design, schemas, pipelines, lakehouse
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
data-platform-engineer | T2 | sonnet | cloud-platforms, lakehouse, data-modeling | lakehouse-architect, pipeline-architect, schema-designer |
genai-architect | T1 | opus | genai, prompt-engineering, ai-data-engineering | — |
kb-architect | T2 | sonnet | — | user |
lakehouse-architect | T2 | sonnet | lakehouse, spark, data-modeling | data-platform-engineer, spark-engineer, schema-designer |
medallion-architect | T1 | sonnet | medallion, data-modeling, lakehouse, data-quality | — |
pipeline-architect | T2 | sonnet | airflow, data-quality, dbt | dbt-specialist, spark-engineer, streaming-engineer |
schema-designer | T2 | sonnet | data-modeling, sql-patterns, data-quality | dbt-specialist, lakehouse-architect, data-quality-analyst, sql-optimizer |
the-planner | T2 | opus | — | user |
AWS, GCP, CI/CD, deployment
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
ai-data-engineer-cloud | T3 | sonnet | gcp, aws, terraform, data-quality, cloud-platforms | ai-data-engineer-gcp, aws-data-architect, user |
ai-data-engineer-gcp | T2 | sonnet | gcp, terraform, cloud-platforms, data-quality | aws-data-architect, user |
ai-prompt-specialist-gcp | T3 | sonnet | prompt-engineering, genai, pydantic, gcp | gcp-data-architect, user |
aws-data-architect | T1 | sonnet | aws, terraform, data-quality | — |
aws-deployer | T3 | sonnet | aws, terraform | aws-lambda-architect, ci-cd-specialist, user |
aws-lambda-architect | T3 | sonnet | aws, terraform | aws-deployer, lambda-builder, user |
ci-cd-specialist | T3 | sonnet | terraform, aws, lakeflow | lambda-builder, aws-lambda-architect, user |
gcp-data-architect | T1 | sonnet | gcp, terraform, cloud-platforms, data-quality | — |
lambda-builder | T3 | sonnet | aws, python, testing | aws-lambda-architect, aws-deployer, user |
supabase-specialist | T3 | opus | supabase, ai-data-engineering, data-modeling | gcp-data-architect, aws-data-architect, user |
dbt, Spark, SQL, Airflow, streaming, data quality
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
ai-data-engineer | T2 | sonnet | ai-data-engineering, data-quality, streaming | pipeline-architect, spark-engineer, streaming-engineer, data-quality-analyst |
airflow-specialist | T3 | sonnet | airflow, sql-patterns, data-quality | spark-engineer, dbt-specialist, streaming-engineer |
dbt-specialist | T2 | sonnet | dbt, data-quality, sql-patterns | schema-designer, spark-engineer, pipeline-architect, data-quality-analyst |
lakeflow-architect | T3 | sonnet | lakeflow, lakehouse, spark, medallion | spark-engineer, dbt-specialist, airflow-specialist |
lakeflow-expert | T3 | sonnet | lakeflow, lakehouse, data-quality, medallion | spark-engineer, airflow-specialist, schema-designer |
lakeflow-pipeline-builder | T3 | sonnet | lakeflow, lakehouse, data-quality, medallion | spark-engineer, airflow-specialist, schema-designer |
lakeflow-specialist | T1 | sonnet | lakeflow, lakehouse, spark, data-quality | — |
qdrant-specialist | T3 | opus | ai-data-engineering, genai | ai-data-engineer, spark-engineer, ai-data-engineer |
spark-engineer | T2 | sonnet | spark, sql-patterns, streaming | pipeline-architect, dbt-specialist, lakehouse-architect |
spark-performance-analyzer | T1 | sonnet | spark, cloud-platforms, lakehouse | — |
spark-specialist | T2 | opus | spark, sql-patterns, cloud-platforms | pipeline-architect, dbt-specialist, lakehouse-architect |
spark-streaming-architect | T3 | sonnet | spark, streaming, lakehouse | spark-engineer, streaming-engineer, airflow-specialist |
spark-troubleshooter | T1 | sonnet | spark, sql-patterns | — |
sql-optimizer | T2 | sonnet | sql-patterns, data-modeling, dbt | spark-engineer, schema-designer, dbt-specialist |
streaming-engineer | T2 | sonnet | streaming, spark, sql-patterns | pipeline-architect, dbt-specialist, lakehouse-architect, ai-data-engineer |
Codebase exploration, meeting analysis, shell, prompts
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
codebase-explorer | T2 | sonnet | — | python-developer, the-planner |
meeting-analyst | T2 | sonnet | — | the-planner, pipeline-architect |
prompt-crafter | T1 | sonnet | python | — |
shell-script-specialist | T2 | sonnet | — | python-developer, ci-cd-specialist |
Fabric lakehouse, pipelines, AI, security
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
fabric-ai-specialist | T3 | sonnet | microsoft-fabric | user, fabric-security-specialist |
fabric-architect | T3 | opus | microsoft-fabric | user, fabric-security-specialist |
fabric-cicd-specialist | T3 | sonnet | microsoft-fabric | user, fabric-security-specialist |
fabric-logging-specialist | T3 | sonnet | microsoft-fabric | user, fabric-security-specialist |
fabric-pipeline-developer | T3 | sonnet | microsoft-fabric | user, fabric-architect |
fabric-security-specialist | T3 | opus | microsoft-fabric | user, user |
Python dev, review, cleanup, documentation, LLM prompts
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
ai-prompt-specialist | T1 | sonnet | prompt-engineering, pydantic, genai | — |
code-cleaner | T2 | sonnet | python | — |
code-documenter | T2 | sonnet | python | — |
code-reviewer | T2 | sonnet | data-quality, sql-patterns, dbt | — |
llm-specialist | T3 | opus | prompt-engineering, pydantic, genai | — |
python-developer | T1 | sonnet | python, pydantic, testing | — |
pytest, data quality, ODCS contracts
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
data-contracts-engineer | T2 | sonnet | data-quality, data-modeling | data-quality-analyst, schema-designer, dbt-specialist |
data-quality-analyst | T2 | sonnet | data-quality, dbt, data-modeling | dbt-specialist, schema-designer, data-contracts-engineer |
test-generator | T2 | sonnet | data-quality, dbt, testing | schema-designer, dbt-specialist, data-quality-analyst |
Brainstorm, Define, Design, Build, Ship, Iterate
| Agent | Tier | Model | KB Domains | Escalates To |
|---|---|---|---|---|
brainstorm-agent | T2 | sonnet | — | define-agent |
build-agent | T2 | opus | — | design-agent |
define-agent | T2 | sonnet | — | design-agent |
design-agent | T2 | opus | — | build-agent |
iterate-agent | T2 | sonnet | — | define-agent, design-agent, build-agent |
ship-agent | T2 | sonnet | — | build-agent |
Which agents know which domain. Use this when the user names a technology.
| KB Domain | Agents |
|---|---|
ai-data-engineering | ai-data-engineer, genai-architect, qdrant-specialist, supabase-specialist |
airflow | airflow-specialist, pipeline-architect |
aws | ai-data-engineer-cloud, aws-data-architect, aws-deployer, aws-lambda-architect, ci-cd-specialist, lambda-builder |
cloud-platforms | ai-data-engineer-cloud, ai-data-engineer-gcp, data-platform-engineer, gcp-data-architect, spark-performance-analyzer, spark-specialist |
data-modeling | data-contracts-engineer, data-platform-engineer, data-quality-analyst, lakehouse-architect, medallion-architect, schema-designer, sql-optimizer, supabase-specialist |
data-quality | ai-data-engineer, ai-data-engineer-cloud, ai-data-engineer-gcp, airflow-specialist, aws-data-architect, code-reviewer, data-contracts-engineer, data-quality-analyst, dbt-specialist, gcp-data-architect, lakeflow-expert, lakeflow-pipeline-builder, lakeflow-specialist, medallion-architect, pipeline-architect, schema-designer, test-generator |
dbt | code-reviewer, data-quality-analyst, dbt-specialist, pipeline-architect, sql-optimizer, test-generator |
gcp | ai-data-engineer-cloud, ai-data-engineer-gcp, ai-prompt-specialist-gcp, gcp-data-architect |
genai | ai-prompt-specialist, ai-prompt-specialist-gcp, genai-architect, llm-specialist, qdrant-specialist |
lakeflow | ci-cd-specialist, lakeflow-architect, lakeflow-expert, lakeflow-pipeline-builder, lakeflow-specialist |
lakehouse | data-platform-engineer, lakeflow-architect, lakeflow-expert, lakeflow-pipeline-builder, lakeflow-specialist, lakehouse-architect, medallion-architect, spark-performance-analyzer, spark-streaming-architect |
medallion | lakeflow-architect, lakeflow-expert, lakeflow-pipeline-builder, medallion-architect |
microsoft-fabric | fabric-ai-specialist, fabric-architect, fabric-cicd-specialist, fabric-logging-specialist, fabric-pipeline-developer, fabric-security-specialist |
prompt-engineering | ai-prompt-specialist, ai-prompt-specialist-gcp, genai-architect, llm-specialist |
pydantic | ai-prompt-specialist, ai-prompt-specialist-gcp, llm-specialist, python-developer |
python | code-cleaner, code-documenter, lambda-builder, prompt-crafter, python-developer |
spark | lakeflow-architect, lakeflow-specialist, lakehouse-architect, spark-engineer, spark-performance-analyzer, spark-specialist, spark-streaming-architect, spark-troubleshooter, streaming-engineer |
sql-patterns | airflow-specialist, code-reviewer, dbt-specialist, schema-designer, spark-engineer, spark-specialist, spark-troubleshooter, sql-optimizer, streaming-engineer |
streaming | ai-data-engineer, spark-engineer, spark-streaming-architect, streaming-engineer |
supabase | supabase-specialist |
terraform | ai-data-engineer-cloud, ai-data-engineer-gcp, aws-data-architect, aws-deployer, aws-lambda-architect, ci-cd-specialist, gcp-data-architect |
testing | lambda-builder, python-developer, test-generator |
Single-sentence purpose per agent, derived from frontmatter description.
ai-data-engineer — AI data engineering specialist for RAG pipelines, vector databases, feature stores, and LLMOps.ai-data-engineer-cloud — Expert Data Engineer for cloud architectures and AI pipelines. Uses KB + MCP validation for best practices.ai-data-engineer-gcp — Elite GCP Data Engineering architect for serverless architectures, AI/ML pipelines, and document processing.ai-prompt-specialist — Prompt engineering specialist for LLMs — extraction, structured output, chain-of-thought, few-shot.ai-prompt-specialist-gcp — Elite Prompt Engineering architect for Google Gemini, Vertex AI, and multi-modal document extraction systems. Masters structured extraction, OCR optimization, and production prompt pipelines. Uses KB + MCP validation.airflow-specialist — Apache Airflow 3.0 SME for DAG development, asset-aware scheduling, and event-driven pipelines.aws-data-architect — AWS data architecture specialist for Lambda, S3, Glue, Redshift, MWAA, and serverless data pipelines.aws-deployer — Executes AWS CLI and SAM CLI deployment commands with validation. Uses KB + MCP validation for safe deployments.aws-lambda-architect — Creates SAM templates with embedded least-privilege IAM policies. Uses KB + MCP validation for secure Lambda deployments.brainstorm-agent — Collaborative exploration specialist for clarifying intent and approach (Phase 0).build-agent — Implementation executor with agent delegation (Phase 3).ci-cd-specialist — DevOps expert for Azure DevOps, Terraform, and Databricks Asset Bundles. Builds CI/CD pipelines for Lambda and Lakeflow deployment with multi-environment promotion. Uses KB + MCP validation for production-ready automation.code-cleaner — Python code cleaning specialist for removing noise and applying modern patterns.code-documenter — Documentation specialist for creating comprehensive, production-ready documentation.code-reviewer — Expert code review specialist ensuring quality, security, and maintainability.codebase-explorer — Elite codebase analyst delivering Executive Summaries + Deep Dives.data-contracts-engineer — Data contract specialist for ODCS, SLA enforcement, schema governance, and producer-consumer agreements.data-platform-engineer — Cloud data platform specialist for Snowflake, Databricks, BigQuery, and infrastructure decisions.data-quality-analyst — Data quality specialist for Great Expectations, Soda, dbt tests, data contracts, and observability.dbt-specialist — dbt Core and dbt Cloud specialist for model development, testing, macros, and project management.define-agent — Requirements extraction and validation specialist (Phase 1).design-agent — Architecture and technical specification specialist (Phase 2).fabric-ai-specialist — Expert in Microsoft Fabric AI capabilities - Copilot, ML models, AI Skills, and Azure OpenAI integration.fabric-architect — Strategic Fabric solution architect for end-to-end architectures using KB + MCP validation.fabric-cicd-specialist — Expert in Microsoft Fabric CI/CD, Git integration, and deployment pipelines.fabric-logging-specialist — Expert in Microsoft Fabric logging, monitoring, KQL queries, and observability.fabric-pipeline-developer — Expert in Fabric Data Factory pipelines, orchestration, and ETL workflows.fabric-security-specialist — Expert in Microsoft Fabric security, governance, and compliance.gcp-data-architect — Google Cloud data architecture specialist for BigQuery, Cloud Run, Pub/Sub, GCS, Dataflow, and Vertex AI.genai-architect — GenAI Systems Architect for multi-agent orchestration, agentic workflows, and production AI systems.iterate-agent — Cross-phase document updater with cascade awareness (All Phases).kb-architect — Knowledge base architect for creating validated, structured KB domains.lakeflow-architect — Databricks Lakeflow expert for building Medallion architecture pipelines. Creates Bronze/Silver/Gold layers with DLT. Uses KB + MCP validation.lakeflow-expert — Databricks Lakeflow (DLT) SME for pipeline development, CDC, data quality, and production deployment. Uses KB + MCP validation.lakeflow-pipeline-builder — Builds Databricks Lakeflow (DLT) pipelines for Medallion Architecture. Uses KB + MCP validation for production-ready pipelines.lakeflow-specialist — Databricks Lakeflow (DLT) specialist for declarative pipelines, materialized views, streaming tables, and expectations.lakehouse-architect — Open table format and catalog specialist for Iceberg, Delta Lake, and lakehouse governance.lambda-builder — AWS Lambda expert for Python serverless file processing. Builds S3-triggered Lambda functions with proper error handling, structured logging, and Parquet output. Uses KB + MCP validation for production-ready code.llm-specialist — Prompt engineering specialist and LLM expert. Masters structured prompting, chain-of-thought reasoning, and AI-powered extraction. Uses KB + MCP validation for optimized, production-ready prompts.medallion-architect — Medallion Architecture specialist for Bronze/Silver/Gold layer design and data quality progression.meeting-analyst — Master communication analyst that transforms meetings into structured, actionable documentation.pipeline-architect — Orchestration specialist for Airflow, Dagster, and pipeline design patterns.prompt-crafter — PROMPT.md builder with SDD-lite phases and Agent Matching Engine.python-developer — Python code architect for data engineering systems — clean patterns, dataclasses, type hints, generators.qdrant-specialist — Elite Qdrant vector database specialist for collection management, point operations, payload filtering, search optimization, and RAG pipeline integration.schema-designer — Data modeling specialist for dimensional modeling, Data Vault, SCD types, and schema evolution.shell-script-specialist — Elite shell scripting specialist for building production-grade Bash scripts with best practices, error handling, and cross-platform compatibility.ship-agent — Feature archival and lessons learned specialist (Phase 4).spark-engineer — PySpark and Spark SQL specialist for distributed data processing at scale.spark-performance-analyzer — Spark performance optimization specialist for tuning memory, partitioning, joins, and I/O.spark-specialist — Apache Spark SME for performance optimization, architecture design, and troubleshooting.spark-streaming-architect — Spark Structured Streaming expert for real-time pipelines, Kafka integration, and stream processing. Uses KB + MCP validation.spark-troubleshooter — Spark debugging specialist for diagnosing OOM errors, data skew, shuffle failures, and job hangs.sql-optimizer — Cross-dialect SQL optimization specialist for query plans, window functions, and performance tuning.streaming-engineer — Stream processing specialist for Flink, Kafka, Spark Streaming, RisingWave, and CDC pipelines.supabase-specialist — Elite Supabase specialist for pgvector, RLS, Edge Functions, Auth, Realtime, and database design.test-generator — Test automation expert for Python. Generates pytest unit tests, integration tests, and fixtures.the-planner — Strategic AI architect that creates comprehensive implementation plans.Cost-optimize by matching task complexity to model capability.
| Model | Share | Use For |
|---|---|---|
| Haiku | ~70% | File exploration, pattern matching, documentation lookup, simple code generation |
| Sonnet | ~20% | Code review, feature implementation, refactoring, API development, most T1/T2 agents |
| Opus | ~10% | Architectural decisions, complex system design, security reviews, T3 agents |
Override rules:
model: wins over task-complexity heuristics.Parallel (independent work, different files):
dbt-specialist + test-generatorcode-reviewer + data-quality-analystschema-designer + pipeline-architectSerial (output feeds next step):
schema-designer → dbt-specialistpipeline-architect → airflow-specialistdefine-agent → design-agent → build-agentBackground (non-blocking):
codebase-explorer, code-documenter, kb-architectkb_domains implies file types (e.g., dbt → models/**/*.sql).description is the semantic anchor.category scopes the match to the right domain.escalation_rules.target provides the handoff graph.To change routing, edit the agent's frontmatter, not this file. Then run:
python3 scripts/generate-agent-router.py