Mantic.sh
Summary
Mantic is a context-aware code search engine that prioritizes relevance over raw speed. After testing across 5 repositories (cal.com, next.js, tensorflow, supabase, chromium), it demonstrates superior result quality compared to grep/ripgrep, despite some trade-offs in speed for very large codebases.
Overall Assessment: 4/5 - Excellent for AI agents, good for developers, needs speed optimization for 100K+ file repos.
What's New in v1.0.25
Enterprise-Grade Context Infrastructure
- Semantic Reranking (Hybrid Intelligence): Combines heuristic speed with neural understanding. Uses local embeddings (
transformers.js) to find "conceptually relevant" code even without exact keyword matches.- Usage:
mantic "verify user" --semantic
- Usage:
- Code Intelligence: Deep understanding of your codebase structure using Tree-sitter.
- Go to Definition:
mantic goto UserServicereturns the exact line number across your entire monorepo. - Find References:
mantic references handleLoginfinds every usage, respecting.gitignore.
- Go to Definition:
- Learned Context (Team Memories): Mantic now remembers which files solved previous queries. These patterns are saved locally (
.mantic/search-patterns.json) and can be committed to git to share knowledge across your team. - Python Support: Now includes first-class support for Python imports in the dependency graph.
- Security & Stability:
- Regex DoS protection for user inputs.
- Command injection mitigations for VS Code extension.
- Safe fallback for non-git directories (scans allow-listed extensions).
Performance Update: v1.0.25 is ~2x faster than previous versions, scanning Chromium (481K files) in <2 seconds.
Tested on 481K files (Chromium) with 100% multi-repo accuracy.
See the CHANGELOG for detailed release notes.
Table of Contents
- About the Project
- Proprietary vs Mantic
- Performance Benchmarks
- Accuracy & Relevance
- Feature Comparison
- Use Case Recommendations
- Installation
- Usage
- Agent Rules
- How It Works
- License
About the Project
Mantic is an infrastructure layer designed to remove unnecessary context retrieval overhead for AI agents. It infers intent from file structure and metadata rather than brute-force reading content, enabling retrieval speeds faster than human reaction time.
Key Benefits
- Speed: Retrieval is consistently under 500ms for most repos, under 4s for massive monorepos (Chromium).
- Efficiency: Reduces token usage by up to 63% by filtering irrelevant files before reading.
- Privacy: Runs entirely locally with zero data egress.
Proprietary vs Mantic (Cost Analysis)
For a team of 100 developers performing 100 searches per day (approx. 3M searches/year):
| Tool | Annual Cost (Est.) | Per-Search Cost | Privacy | |------|---|---|---| | Mantic | $0 | $0 | Local-First | | Vector Embeddings (DIY) | $1,680 - $10,950* | $0.0005 - $0.003 | Cloud | | SaaS Alternatives | $46,800+ | $0.015+ | Cloud |
Note: Mantic costs are zero. Vector/SaaS costs are estimates based on standard managed infrastructure (e.g. Pinecone/Weaviate managed pods + compute) or per-seat Enterprise licensing (e.g. GitHub Copilot Enterprise).
Performance Benchmarks
Speed Comparison (Real-world queries)
| Repository | Files | Query | Mantic v1.0.25 | ripgrep | fzf | Verdict | |------------|-------|-------|-----------------|---------|-----|--------| | cal.com | 9.7K | "stripe payment" | 0.288s | 0.121s | 0.534s | Fast | | next.js | 25K | "router server" | 0.440s | 0.034s | 0.049s | Fast | | tensorflow | 35K | "gpu" | 0.550s | 0.022s | N/A | Fast | | chromium | 481K | "ScriptController" | 1.961s | 0.380s | 0.336s | <2s (Massive) |
Speed Verdict:
- Caching works well for most repos (4-17% improvement on second run).
- Large repos (Chromium) show modest but consistent improvements.
- Mantic is slower than ripgrep/fzf on raw speed but prioritizes ranking.
Accuracy & Relevance Analysis
Major Strengths
1. Exact Path Matching
- Query:
"router server"in next.js - Mantic: Found
packages/next/src/server/lib/router-server.ts(Score: 220) - ripgrep: Found files mentioning "router" and "server" separately (many false positives)
- Verdict: Mantic found the exact file that matches the intent.
2. CamelCase Detection
- Query:
"ScriptController"in chromium - Mantic: Found
script_controller.h,script_controller.cc(Score: 200) - ripgrep: Requires manual regex
script.*controller - Verdict: Mantic's CamelCase detection is production-ready.
3. Directory Boosting for Acronyms
- Query:
"gpu"in tensorflow - Mantic: Prioritized files in
tensorflow/lite/delegates/gpu/ - Verdict: Mantic correctly prioritizes structural relevance.
4. Path Sequence Matching
- Query:
"blink renderer core dom"in chromium - Mantic: Found
third_party/blink/renderer/core/dom/README.md - Verdict: Mantic matches multi-term path queries perfectly.
Feature Comparison Matrix
| Feature | Mantic | ripgrep | ag | fzf | |---------|--------|---------|----|-----| | Text Search Speed | 2-10x slower | Fastest | Slow (large repos) | Very Fast | | Relevance Ranking | Excellent | None | None | Basic | | Path Structure Awareness | Perfect | None | None | Partial | | CamelCase Detection | Yes | No | No | No | | Exact Filename Matching | Yes | No | No | Yes | | Multi-Word Queries | Semantic | Regex needed | Regex needed | AND logic | | Go to Definition | Yes (Cross-Repo) | No | No | No | | Find References | Yes | No | No | No | | Impact Analysis | Yes | No | No | No | | Zero-Query Mode | Yes | No | No | No |
Use Case Recommendations
Best For
- AI Agents (Context-aware search with metadata)
- Finding Files by Intent ("Where is payment code?")
- Understanding Code Structure (Path sequence queries)
- Code Reviews (Impact analysis shows blast radius)
Not Ideal For
- Quick Text Searches ("Find all TODOs" -> Use ripgrep)
- Very Large Repos (100K+) (Speed tradeoff: 4s vs 0.3s)
- Exact String Matching (Use ripgrep with -F)
- Interactive File Browsing (Use fzf)
Installation
CLI Installation
Quick Start (no installation required):
npx mantic.sh@latest "your search query"
New Commands:
# Semantic Search (Neural Reranking)
npx mantic.sh@latest "verify user identity" --semantic
# Go to Definition
npx mantic.sh@latest goto "UserService"
# Find References
npx mantic.sh@latest references "handleLogin"
From Source:
git clone https://github.com/marcoaapfortes/Mantic.sh.git
cd Mantic.sh
npm install
npm run build
npm link
MCP Server Installation
Mantic works as an MCP (Model Context Protocol) server for Claude Desktop, Cursor, VS Code, and other MCP-compatible tools.
One-Click Install:
Manual Configuration (for Claude Desktop or other MCP clients):
Add this to your MCP settings file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"mantic": {
"command": "npx",
"args": ["-y", "mantic.sh@latest", "server"]
}
}
}
Usage
Basic Search
Find files matching your intent:
mantic "stripe payment integration"
Returns JSON with ranked files, confidence scores, and token estimates.
Advanced Usage
Zero-Query Mode (proactive context):
mantic ""
# Shows modified files, related dependencies, impact analysis
Context Carryover (session mode):
mantic "auth logic" --session my-task
# Previously viewed files get +150 boost
Impact Analysis:
mantic "payment processing" --impact
# Shows blast radius and depe