KiroGraph gives Kiro (and 34 other MCP-capable tools) a semantic knowledge graph that’s pre-indexed and always up to date. Instead of scanning files to understand your code, your AI agent queries the graph instantly — symbol relationships, call graphs, type hierarchies, and impact radius — all in a single MCP tool call.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/davide-desio-eleva/kirograph/llms.txt
Use this file to discover all available pages before exploring further.
Quickstart
Install KiroGraph and wire it up to Kiro in under 5 minutes.
How It Works
Understand the indexing layers, graph model, and semantic engines.
MCP Tools Reference
Full reference for all 126 MCP tools across every module.
CLI Reference
Every CLI command, flag, and option with real examples.
Why KiroGraph?
When you ask Kiro to work on a complex task, it explores your codebase using file reads, grep, and glob searches. Every one of those is a tool call — and tool calls consume context and slow things down. KiroGraph replaces that entire loop with a pre-built graph stored locally in SQLite. One call tokirograph_context returns entry points, related symbols, call edges, and code snippets — often everything the agent needs to understand the task without reading a single file.
Fewer Tool Calls
One
kirograph_context call replaces 10–30 file reads, grep searches, and glob scans.100% Local
All data lives in
.kirograph/ on your machine. No API keys, no external services, nothing leaves your network.33+ Languages
tree-sitter AST parsing across TypeScript, Python, Go, Rust, Java, C#, Ruby, and 26 more.
Core Modules
Security Analysis
Reachability-aware CVE scanning — know whether vulnerable code is actually called from your entry points.
Architecture Analysis
Package graph, layer detection, and coupling metrics (Ca/Ce/instability) with snapshot diffing.
Persistent Memory
Cross-session observations linked to code symbols — decisions, errors, and patterns that survive context resets.
Semantic Search
9 pluggable engines including TurboQuant (20–30× embedding compression) and TurboVec (Rust/SIMD).
LLM Wiki
Karpathy-style knowledge wiki that compounds across sessions with conflict resolution and FTS search.
Token Optimization
Shell compression (60–90% savings), Caveman prose compression, and cached file reads.
Integrations
KiroGraph has native support for Kiro and experimental support for 34 other MCP-capable tools including Cursor, Claude Code, GitHub Copilot, Windsurf, Cline, Codex, and more.View all integrations
See the full list of supported tools and how auto-detection works.
