Comparative Analysis Workflow

Comparative Analysis Workflow

CONCEPT:RES-001

Parallel execution workflow for comparative analysis and innovation extraction using the Unified Parallel Engine

Steps

Step 1: Pre Flight Config

Agent: classification-agent Tools: rep_rm_workspace, rep_rm_projects

Gather inputs, configuration params, target codebases, papers, concept registries, and C4 files from the user. Expected: pre_flight_config_artifacts

Step 2: Discovery Classification [depends_on: pre_flight_config]

Agent: classification-agent Tools: rep_rm_workspace, rep_rm_projects

Classify input files (codebases vs research), detect language ecosystems, and calculate composite relevance rankings. Expected: discovery_classification_artifacts

Step 3: Codebase Audit [depends_on: discovery_classification]

Agent: audit-agent Tools: rep_rm_projects

Analyze target codebase governance, active ecosystem health, CI posture, and basic complexity metrics. Expected: codebase_audit_artifacts

Step 4: Architecture Discovery [depends_on: discovery_classification]

Agent: audit-agent Tools: rep_rm_projects

Identify existing C4 diagrams or auto-generate AST-based C4 models, tracing module imports and hot paths. Expected: architecture_discovery_artifacts

Step 5: Security Reliability Check [depends_on: discovery_classification]

Agent: security-agent Tools: rep_rm_projects

Scan for OWASP Top 10 patterns, CWE exposures, hardcoded keys, and evaluate test coverage/pyramid intent. Expected: security_reliability_check_artifacts

Step 6: Documentation DX Review [depends_on: discovery_classification]

Agent: audit-agent Tools: rep_rm_projects

Evaluate readme grades, docstring coverage, concurrency, and performance benchmark suites. Expected: documentation_dx_review_artifacts

Step 7: Innovation Extraction [depends_on: discovery_classification]

Agent: innovation-agent Tools: rep_rm_workspace

Extract innovation claims, biomimicry mappings, and TRIZ inventive principles from papers and code. Expected: innovation_extraction_artifacts

Step 8: Concept Cross Reference [depends_on: discovery_classification]

Agent: innovation-agent Tools: rep_rm_workspace

Perform concept-ID seeded cross-referencing against registries to map findings to unique architectural concept IDs. Expected: concept_cross_reference_artifacts

Step 9: Architecture Gap Analysis [depends_on: codebase_audit, architecture_discovery, innovation_extraction]

Agent: synthesis-agent Tools: rep_rm_projects

Perform architecture topology differential and gap analysis between targets to discover wiring opportunities. Expected: architecture_gap_analysis_artifacts

Step 10: Generate Comparison Report [depends_on: security_reliability_check, documentation_dx_review, concept_cross_reference, architecture_gap_analysis]

Agent: synthesis-agent Tools: rep_rm_workspace

Compile all intermediate domain analysis results into a single unified Markdown comparative analysis report. Expected: generate_comparison_report_artifacts

Step 11: KG Persistence Cleanup [depends_on: generate_comparison_report]

Agent: synthesis-agent Tools: graph_write

Ingest final reports, wiring audits, and relationship edges into Graph-OS Knowledge Graph, and purge temporary file paths. Expected: kg_persistence_cleanup_artifacts

Output

• Graded comparative analysis report persisted in .specify/
• Standardized 0-100 scorecards across all target domains with radar chart inputs
• Nodes and edges persisted to Knowledge Graph representing cross-domain/cross-project mappings

Execution

Run this workflow as a dependency-ordered DAG. Steps with no unmet depends_on run in parallel; dependents run after their prerequisites complete.

• Run first (in parallel): Step 1 — Pre Flight Config
• After level 0: Step 2 — Discovery Classification
• After level 1: Step 3 — Codebase Audit; Step 4 — Architecture Discovery; Step 5 — Security Reliability Check; Step 6 — Documentation DX Review; Step 7 — Innovation Extraction; Step 8 — Concept Cross Reference
• After level 2: Step 9 — Architecture Gap Analysis
• After level 3: Step 10 — Generate Comparison Report
• After level 4: Step 11 — KG Persistence Cleanup

Execution: If graph-os is reachable, offload the whole DAG via graph_orchestrate action=execute_workflow (or the kg-delegate skill) for true parallel/swarm execution. Otherwise execute the steps natively in dependency order: run steps with no unmet depends_on in parallel, then their dependents.