How ga4gh/ga4gh_workflow_clinic_gsoc_2026 works
Workflow Clinic analyzes scientific workflow files by parsing them into a language-neutral WorkflowBundle, running portability rules and optional AI review, then producing a diagnosis report. A separate repair command loads findings and delegates proposed fixes to the Doctor cascade; publishing can turn findings into GitHub issues. The README also describes Snakemake support and metadata/structure validation as planned capabilities, but the supplied tree currently shows a Nextflow parser and only some rule implementations. Edges reflect the README or SOURCE INDEX references; unsampled implementation details are not assumed.
Main entry points: coordinator.py Examine Coordinator · cli.py CLI Commands · agent.py AI Critic · fix_coordinator.py Fix Coordinator
Components 22
CLI and Coordination
- CLI Commands
src/workflow_clinic/cli.py - Examine Coordinator
src/workflow_clinic/services/coordinator.py - Fix Coordinator
src/workflow_clinic/services/fix_coordinator.py
Workflow Analysis
- Parser Registry
src/workflow_clinic/parsers/registry.py - Nextflow Parser
src/workflow_clinic/parsers/nextflow.py - WorkflowBundle
src/workflow_clinic/models/workflow_bundle.py - Rule Engine
src/workflow_clinic/rules - Diagnosis Models
src/workflow_clinic/models/diagnosis.py - Diagnosis Report
src/workflow_clinic/models/diagnosis.py
AI Review
- AI Critic
src/workflow_clinic/critic/agent.py - Remediation Knowledge
src/workflow_clinic/advisor/retriever.py - LLM Utilities
src/workflow_clinic/utils/llm.py
Repair
- Doctor Runner
src/workflow_clinic/doctor/runner.py - Fixer Strategies
src/workflow_clinic/doctor/fixers - File Patcher
src/workflow_clinic/doctor/patcher.py - Fix Proposals
src/workflow_clinic/models/fix.py
Reporting
- Publish Coordinator
src/workflow_clinic/services/publish_coordinator.py - Issue Generator
src/workflow_clinic/reporting/issue_generator.py - GitHub Publisher
src/workflow_clinic/reporting/github_publisher.py
Other components
- Workflow Author
- LLM Provider
- GitHub
Connections 25
- Workflow Author → CLI Commands: runs commands
README.md - CLI Commands → Examine Coordinator: starts examination
src/workflow_clinic/cli.py - CLI Commands → Fix Coordinator: starts repair
src/workflow_clinic/cli.py - CLI Commands → Publish Coordinator: starts publishing
src/workflow_clinic/cli.py - Examine Coordinator → Parser Registry: selects parser
src/workflow_clinic/services/coordinator.py - Parser Registry → Nextflow Parser: dispatches parsing
- Nextflow Parser → WorkflowBundle: builds model
src/workflow_clinic/parsers/nextflow.py - Examine Coordinator → WorkflowBundle: analyzes workflow
README.md - Examine Coordinator → Rule Engine: runs rules
src/workflow_clinic/services/coordinator.py - Examine Coordinator → AI Critic: runs optional audit
src/workflow_clinic/services/coordinator.py - AI Critic → Remediation Knowledge: retrieves guidance
src/workflow_clinic/critic/agent.py - AI Critic → LLM Utilities: uses model utilities
src/workflow_clinic/critic/agent.py - LLM Utilities → LLM Provider: resolves provider
- AI Critic → Diagnosis Models: enriches findings
src/workflow_clinic/critic/agent.py - Examine Coordinator → Diagnosis Models: builds report
src/workflow_clinic/services/coordinator.py - Examine Coordinator → Diagnosis Report: saves report
src/workflow_clinic/services/coordinator.py - Fix Coordinator → Diagnosis Report: loads findings
src/workflow_clinic/services/fix_coordinator.py - Fix Coordinator → Doctor Runner: runs cascade
src/workflow_clinic/services/fix_coordinator.py - Doctor Runner → Fixer Strategies: selects strategies
src/workflow_clinic/doctor/runner.py - Doctor Runner → Fix Proposals: records proposals
src/workflow_clinic/doctor/runner.py - Fixer Strategies → File Patcher: may delegate patching
src/workflow_clinic/doctor/fixers/paths.py - Publish Coordinator → Issue Generator: generates issues
src/workflow_clinic/services/publish_coordinator.py - Publish Coordinator → GitHub Publisher: publishes issues
src/workflow_clinic/services/publish_coordinator.py - GitHub Publisher → Issue Generator: uses issue payloads
src/workflow_clinic/reporting/github_publisher.py - GitHub Publisher → GitHub: creates issues
src/workflow_clinic/reporting/github_publisher.py