`get_impact_radius` vs `get_review_context` in code-review-graph: Functional Differences Explained

get_impact_radius performs pure graph analysis to identify affected code nodes, while get_review_context builds on that analysis to deliver a complete review package with source snippets and actionable guidance.

Both tools are part of the code-review-graph knowledge-graph suite for automated code review. They share underlying graph traversal logic but serve fundamentally different purposes in a review pipeline. Understanding when to use each tool helps you build efficient CI workflows and reviewer-friendly interfaces.

Core Purpose: Graph Analysis vs. Review Package Assembly

The primary functional difference lies in their scopes:

• get_impact_radius — Computes the blast radius of code changes. It answers "what breaks?" by traversing dependency edges from changed files to identify impacted nodes.
• get_review_context — Assembles a token-efficient review payload. It answers "what should a reviewer focus on?" by wrapping impact analysis with source excerpts, risk scoring, and actionable hints.

This distinction makes get_impact_radius ideal for downstream automation, while get_review_context targets human reviewers or LLM-based review agents.

Source Implementation and Data Flow

get_impact_radius in code_review_graph/tools/query.py

The core implementation resides in lines 107–119 of query.py. The tool calls the graph store's native get_impact_radius method and returns raw structural data:

# From code_review_graph/tools/query.py (lines 107-119)

def get_impact_radius(
    changed_files: List[str],
    max_depth: int = 2,
    detail_level: str = "standard",
    # ... additional params

) -> Dict[str, Any]:
    # Direct graph store invocation

    result = store.get_impact_radius(changed_files, max_depth=max_depth)
    # Returns: changed_nodes, impacted_nodes, edges, summary fields

The tool exposes one control parameter: max_depth (default 2) limits BFS traversal hops through the dependency graph.

get_review_context in code_review_graph/tools/review.py

The implementation spans lines 26–119 of review.py. This tool calls get_impact_radius internally then layers on additional processing:

# From code_review_graph/tools/review.py (lines 26-119)

def get_review_context(
    changed_files: List[str],
    max_depth: int = 2,
    include_source: bool = True,
    max_lines_per_file: int = 150,
    detail_level: str = "standard",
    # ... additional params

) -> Dict[str, Any]:
    # Step 1: Obtain impact radius via store

    impact = store.get_impact_radius(changed_files, max_depth=max_depth)
    
    # Step 2: Estimate token cost

    token_estimate = estimate_tokens(impact)
    
    # Step 3: Extract source snippets (if enabled)

    snippets = extract_source_snippets(changed_files, max_lines_per_file)
    
    # Step 4: Generate review guidance

    guidance = generate_review_guidance(impact, test_coverage_data)
    
    # Returns: wrapped context with risk, test_gaps, next_tool_suggestions

Additional control parameters in get_review_context:

• include_source — Toggle source snippet extraction
• max_lines_per_file — Cap snippet size for token efficiency

Return Structure Comparison

The output schemas reveal the functional gap between these tools:

get_impact_radius Output

{
  "status": "ok",
  "summary": "Blast radius for 2 changed file(s): 3 functions, 5 impacted dependencies",
  "changed_files": ["src/auth.py", "src/api.py"],
  "changed_nodes": [
    {"id": "auth.login", "type": "function", "file": "src/auth.py"},
    {"id": "api.validate_token", "type": "function", "file": "src/api.py"}
  ],
  "impacted_nodes": [
    {"id": "middleware.auth_check", "type": "function", "file": "src/middleware.py"}
  ],
  "impacted_files": ["src/middleware.py", "src/models/user.py"],
  "edges": [
    {"source": "auth.login", "target": "middleware.auth_check", "type": "calls"}
  ],
  "truncated": false,
  "total_impacted": 5
}

get_review_context Output

{
  "status": "ok",
  "summary": "Review context for 2 changed file(s): medium risk, 1 test gap identified",
  "context": {
    "changed_files": ["src/auth.py", "src/api.py"],
    "impacted_files": ["src/middleware.py", "src/models/user.py"],
    "graph": {
      "changed_nodes": [...],
      "impacted_nodes": [...],
      "edges": [...]
    },
    "source_snippets": {
      "src/auth.py": "def login(username, password):\n    # 15 lines...",

      "src/api.py": "def validate_token(token):\n    # 23 lines..."

    },
    "review_guidance": "1 changed function(s) lack test coverage: login. Changes impact 2 other files. Consider splitting auth logic."
  },
  "risk": "medium",
  "test_gaps": 1,
  "next_tool_suggestions": ["get_test_recommendations", "get_security_audit"]
}

Use-Case Mapping: When to Use Each Tool

Scenario	Recommended Tool	Rationale
CI impact gates / downstream analysis	get_impact_radius	Lightweight, no source extraction overhead
Human code review interfaces	get_review_context	Includes readable snippets and guidance
LLM-based review agents	get_review_context	Pre-packaged context reduces token costs
Dependency audit dashboards	get_impact_radius	Raw graph data for custom visualization
Automated risk scoring pipelines	Either	get_impact_radius for custom scoring; get_review_context for built-in risk levels

Detail Level Behavior: Minimal Mode Differences

Both tools support detail_level="minimal", but the interpretation differs:

get_impact_radius with detail_level="minimal"

• Returns concise summary with risk-scored counts
• Omits full node and edge lists
• Keeps structural fields (changed_files, impacted_files)

get_review_context with detail_level="minimal"

• Drops source snippets entirely regardless of include_source setting
• Returns high-level risk, file counts, key entities, and next_tool_suggestions
• Most compact payload for token-constrained environments

# Minimal mode comparison

from code_review_graph.tools import get_impact_radius, get_review_context

# Impact radius: graph summary only

impact = get_impact_radius(
    changed_files=["src/auth.py"],
    detail_level="minimal"
)

# Contains: summary, total_impacted, changed_files count

# Review context: guidance without snippets

review = get_review_context(
    changed_files=["src/auth.py"],
    include_source=True,  # Ignored in minimal mode

    detail_level="minimal"
)

# Contains: summary, risk level, test_gaps, next_tool_suggestions

# Missing: context.source_snippets

Practical Code Examples

Basic Impact Analysis

from code_review_graph.tools.query import get_impact_radius

result = get_impact_radius(
    changed_files=["src/auth.py", "src/api.py"],
    max_depth=2,
    detail_level="standard"
)

print(f"Affected nodes: {result['total_impacted']}")
print(f"Impact spread: {len(result['impacted_files'])} files")

Complete Review Package

from code_review_graph.tools.review import get_review_context

payload = get_review_context(
    changed_files=["src/auth.py"],
    max_depth=2,
    include_source=True,
    max_lines_per_file=150,
    detail_level="standard"
)

# Access structured guidance

print(f"Risk level: {payload['risk']}")
print(f"Test gaps: {payload['test_gaps']}")

# Iterate suggested follow-up tools

for tool in payload.get('next_tool_suggestions', []):
    print(f"Consider running: {tool}")

Public Entry Points in main.py

Both tools are exposed through CLI/API entry points in code_review_graph/main.py:

• get_impact_radius_tool: lines 19–44
• get_review_context_tool: lines 88–118

Summary

• get_impact_radius in code_review_graph/tools/query.py (lines 107–119) provides raw graph analysis — changed nodes, impacted dependencies, and traversal edges without source content.

• get_review_context in code_review_graph/tools/review.py (lines 26–119) wraps the impact analysis with source snippets, token estimates, risk scoring, and actionable guidance for reviewers.

• Shared foundation: Both use store.get_impact_radius() for graph traversal and respect max_depth and detail_level parameters.

• Key differentiators: get_review_context adds include_source, max_lines_per_file, and produces review_guidance and next_tool_suggestions.

• Minimal mode divergence: get_review_context drops source snippets entirely in minimal mode; get_impact_radius retains core structural data.

Frequently Asked Questions

What is the relationship between get_impact_radius and get_review_context?

get_review_context depends on get_impact_radius. It calls store.get_impact_radius() internally to obtain the base graph analysis, then enriches that data with source extraction and guidance generation. You cannot use get_review_context without the underlying impact analysis logic.

Can I use get_impact_radius directly instead of get_review_context?

Yes. Use get_impact_radius when you need only structural dependency data — for example, to drive custom downstream tools, build visualization dashboards, or implement your own risk scoring. Use get_review_context when you need a complete review payload ready for human or LLM consumption.

Why does get_review_context ignore include_source=True in minimal mode?

The detail_level="minimal" flag is designed for maximum token efficiency. Source snippets are the largest payload component, so they are unconditionally excluded regardless of the include_source parameter. This ensures predictable payload size for CI systems and constrained environments.

Where are these tools registered in the codebase?

Public entry points are defined in code_review_graph/main.py: get_impact_radius_tool (lines 19–44) and get_review_context_tool (lines 88–118). The core implementations are in code_review_graph/tools/query.py and code_review_graph/tools/review.py respectively.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}