dipeo-codegen-pipeline
Router skill for DiPeO code generation pipeline (TypeScript specs → IR → Python/GraphQL). Use when task mentions TypeScript models, IR builders, generated code diagnosis, or codegen workflow. For simple tasks, handle directly; for complex work, escalate to dipeo-codegen-pipeline agent.
When & Why to Use This Skill
This Claude skill serves as an intelligent router and orchestrator for the DiPeO code generation pipeline, bridging TypeScript specifications with Python and GraphQL outputs. It optimizes developer workflows by handling routine lookups, pattern references, and minor model tweaks directly, while seamlessly escalating complex AST transformations and IR builder modifications to specialized agents. By automating the decision-making process between simple tasks and complex engineering workflows, it ensures high-quality code generation and efficient pipeline management.
Use Cases
- Workflow Guidance: Providing instant answers regarding the codegen pipeline structure, IR (Intermediate Representation) schemas, and execution commands.
- Code Diagnosis & Traceability: Tracing issues across the generation stack to identify whether a bug originates in the TypeScript spec, the IR builder, or the Jinja templates.
- Automated Pattern Enforcement: Ensuring naming conventions (such as snake_case) and type mapping rules are strictly followed during minor model updates.
- Intelligent Task Escalation: Automatically routing complex architectural changes, such as creating new node types or modifying type conversion logic, to the dedicated dipeo-codegen-pipeline agent.
- Read-only Pipeline Audits: Reviewing generated code structures and TypeScript model designs to ensure alignment with system architecture without triggering full rebuilds.
| name | dipeo-codegen-pipeline |
|---|---|
| description | Router skill for DiPeO code generation pipeline (TypeScript specs → IR → Python/GraphQL). Use when task mentions TypeScript models, IR builders, generated code diagnosis, or codegen workflow. For simple tasks, handle directly; for complex work, escalate to dipeo-codegen-pipeline agent. |
| allowed-tools | Read, Grep, Glob, Bash, Skill |
DiPeO Codegen Pipeline Router
Domain: Complete TypeScript → IR → Python/GraphQL pipeline (/dipeo/models/src/, /dipeo/infrastructure/codegen/, generated code diagnosis).
Quick Decision: Skill or Agent?
✅ Handle Directly (This Skill)
- Simple lookups: Understanding codegen workflow, reviewing TypeScript specs
- Read-only tasks: Checking generated code structure, reviewing IR output
- Pattern reference: snake_case naming rules, type mapping examples
- Small spec tweaks: Minor TypeScript field changes (<5 lines, 1 file)
- Workflow questions: "How do I run codegen?", "What's the IR structure?"
❌ Escalate to Agent
- New node types: Creating complete TypeScript specs with IR builders
- IR builder changes: Modifying AST processing, type conversion logic
- Generated code diagnosis: Tracing why generated code is wrong
- Template modifications: Changing Jinja templates for code generation
- Complex spec changes: Multi-field changes affecting multiple generated files
- Type system updates: Changes to UnifiedTypeConverter or type mappings
Agent: Task(dipeo-codegen-pipeline, "your detailed task description")
Documentation Sections (Load On-Demand)
Use Skill(doc-lookup) with these anchors when you need detailed context:
Part 1: TypeScript Model Design:
docs/agents/codegen-pipeline.md#your-role-as-model-architect- Model locations and structuredocs/agents/codegen-pipeline.md#type-system-design-principles- CRITICAL: snake_case and type safetydocs/agents/codegen-pipeline.md#workflows- Creating/modifying node types
Part 2: IR Builder System:
docs/agents/codegen-pipeline.md#ir-builder-architecture- Architecture and directory structuredocs/agents/codegen-pipeline.md#pipeline-system- Pipeline, type conversion, AST processing
Part 3: Code Generation:
docs/agents/codegen-pipeline.md#template-system- Templates and generated code structuredocs/agents/codegen-pipeline.md#generation-workflow- Complete make codegen workflow
Part 4: Diagnosis:
docs/agents/codegen-pipeline.md#your-critical-responsibility- Tracing TypeScript → IR → Python
Part 5 & 6: Workflow & Collaboration:
docs/agents/codegen-pipeline.md#complete-workflow- End-to-end steps and validationdocs/agents/codegen-pipeline.md#when-to-engage-other-agents- Escalation paths
Example doc-lookup call:
python .claude/skills/doc-lookup/scripts/section_search.py \
--query "naming-standards" \
--paths docs/agents/codegen-pipeline.md \
--top 1
Escalation to Other Agents
To dipeo-package-maintainer: Runtime handler issues, service architecture (if generated code is correct) To dipeo-backend: GraphQL schema deployment, server config (if generation is correct)
Typical Workflow
- Assess complexity: Simple lookup/guidance vs. complex generation task
- If simple: Load relevant section via
Skill(doc-lookup), provide guidance - If diagnosis needed: Trace TypeScript → IR → Python (use diagnosis docs)
- If complex: Escalate with
Task(dipeo-codegen-pipeline, "task details")