code-review-graph

---

name: code-review-graph description: AST knowledge graph: change review, exploration, debugging, refactoring. Trigger: knowledge graph, code review, impact analysis, blast radius, refactoring.

code-review-graph — Knowledge Graph Toolset

This project connects to an AST-based code knowledge graph via the code-review-graph MCP. When exploring code, prefer graph tools first, then fall back to Grep/Glob/Read.

Token Efficiency Rules (Global)

• Always call get_minimal_context(task="<your task>") first, then use other graph tools.
• Use detail_level="minimal" for all calls; only upgrade to "standard" when necessary.
• Goal: complete the task within 5 tool calls and 800 output tokens.

---

Workflow 1: Change Review

Risk-aware code review using the knowledge graph.

Steps

1. Run detect_changes to get change analysis with risk scores.
2. Run get_affected_flows to find affected execution paths.
3. For high-risk functions, run query_graph pattern="tests_for" to check test coverage.
4. Run get_impact_radius to understand the scope of impact.
5. For changes without test coverage, suggest specific test cases.

Output Format

Grouped by risk level (high/medium/low), including: what changed, test coverage status, improvement suggestions, merge recommendation.

---

Workflow 2: Codebase Exploration

Quickly understand codebase structure using the graph.

Steps

1. Run list_graph_stats to view overall metrics.
2. Run get_architecture_overview to understand high-level module structure.
3. Use list_communities to find main modules, use get_community for details.
4. Use semantic_search_nodes to find functions/classes by name or keyword.
5. Use query_graph's callers_of, callees_of, imports_of to trace relationships.
6. Use list_flows and get_flow to understand execution paths.

Tips

• Start broad (stats, architecture), then narrow to specific areas.
• children_of to see all functions and classes in a file; find_large_functions to locate complex code.

---

Workflow 3: Issue Debugging

Systematically trace and debug issues using the graph.

Steps

1. Use semantic_search_nodes to find code related to the issue.
2. Use query_graph's callers_of and callees_of to trace call chains.
3. Use get_flow to view the complete execution path of suspicious areas.
4. Run detect_changes to check if recent changes triggered the issue.
5. Use get_impact_radius on suspicious files to see the affected scope.

Tips

• Check both callers and callees to understand the full context.
• View affected execution flows to find the entry point that triggers the bug.
• Recent changes are the most common source of new issues.

---

Workflow 4: Safe Refactoring

Plan and execute refactoring confidently using the graph.

Steps

1. Use refactor_tool mode="suggest" to get refactoring suggestions.
2. Use refactor_tool mode="dead_code" to find dead code.
3. For renaming, use refactor_tool mode="rename" to preview all affected locations.
4. Use apply_refactor_tool with refactor_id to apply the rename.
5. After changes, run detect_changes to verify impact.

Safety Checks

• Always preview before applying (rename mode shows a list of edits).
• Check get_impact_radius before large-scale refactoring.
• Use get_affected_flows to ensure critical paths are not broken.
• find_large_functions to identify large functions that need splitting.