Code Enhancer Workflow

Code Enhancer Workflow

CONCEPT:DEV-002

Parallel execution workflow for code enhancement and multi-domain deep-dive auditing using the Unified Parallel Engine

Steps

Step 1: Detect Language

Agent: analysis-agent Tools: rep_rm_workspace, rep_rm_projects

Detect primary/secondary languages, build systems, available linters, and test frameworks in the target project. Expected: detect_language_artifacts

Step 2: Project Analysis [depends_on: detect_language]

Agent: analysis-agent Tools: rep_rm_workspace, rep_rm_projects

Scan codebase for architectural patterns, libraries, Pydantic-AI agent setups, externalized prompts, and dependency freshness. Expected: project_analysis_artifacts

Step 3: Run Linters [depends_on: detect_language]

Agent: analysis-agent Tools: rep_rm_projects

Orchestrate pre-commit hooks and language-appropriate linters (ruff, mypy, bandit, eslint, go vet) to scan for code issues. Expected: run_linters_artifacts

Step 4: Run Tests [depends_on: detect_language]

Agent: test-agent Tools: rep_rm_projects

Locate and execute the project test suite with automated test framework detection and timeout protections. Expected: run_tests_artifacts

Step 5: Deep Code Analysis [depends_on: project_analysis]

Agent: analysis-agent Tools: rep_rm_projects

Examine module coupling, cyclomatic complexity, directory density, code duplication, and smell patterns. Expected: deep_code_analysis_artifacts

Step 6: Security Analysis [depends_on: project_analysis]

Agent: security-agent Tools: rep_rm_projects

Evaluate attack surface, credential exposure, CWE vulnerability patterns, threat model alignments, and API/SSO security controls. Expected: security_analysis_artifacts

Step 7: Documentation Audit [depends_on: project_analysis]

Agent: doc-agent Tools: rep_rm_workspace

Audit repository documentation (README.md, AGENTS.md, docs/) for quality, completeness, Keep a Changelog standard compliance, and drift. Expected: documentation_audit_artifacts

Step 8: Concept Traceability Audit [depends_on: project_analysis]

Agent: doc-agent Tools: rep_rm_workspace

Scan codebase docstrings, test suites, and documentation files for CONCEPT ID annotations and map traceability/parity. Expected: concept_traceability_audit_artifacts

Step 9: User Interface Analysis [depends_on: project_analysis]

Agent: analysis-agent Tools: rep_rm_workspace

Evaluate web or terminal interface files against Usability Heuristics and WCAG accessibility standards. Expected: user_interface_analysis_artifacts

Step 10: Generate Report [depends_on: run_linters, run_tests, deep_code_analysis, security_analysis, documentation_audit, concept_traceability_audit, user_interface_analysis]

Agent: synthesis-agent Tools: rep_rm_workspace

Consolidate the multi-domain findings into a prettified 0-100 graded report with prioritized TODOs and structured SDD handoffs. Expected: generate_report_artifacts

Step 11: KG Persistence [depends_on: generate_report]

Agent: synthesis-agent Tools: graph_write

Persist the generated report, grading metrics, and identified conceptual/code linkages in Graph-OS Knowledge Graph. Expected: kg_persistence_artifacts

Output

• Unified multi-domain codebase enhancement audit report in .specify/
• Prioritized TODO checklist and SDD-ready implementation specification
• Nodes and edges persisted to Knowledge Graph representing code quality and compliance status

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 — Detect Language
• After level 0: Step 2 — Project Analysis; Step 3 — Run Linters; Step 4 — Run Tests
• After level 1: Step 5 — Deep Code Analysis; Step 6 — Security Analysis; Step 7 — Documentation Audit; Step 8 — Concept Traceability Audit; Step 9 — User Interface Analysis
• After level 2: Step 10 — Generate Report
• After level 3: Step 11 — KG Persistence

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.