Power your agent
คำสั่งของผู้เผยแพร่เอง ไม่ใช่เอนด์พอยต์ของ PlugMCP
Host a copy with our keys and logs.
npx -y [email protected]Claude Code ไม่มีลิงก์ติดตั้งสำหรับคำสั่งแบบโลคัล ให้คัดลอกคำสั่งด้านล่างแล้วรันในเทอร์มินัล
claude mcp add io.github.jghiringhelli-codeseeker -- npx -y codeseeker@2.1.2รันในเทอร์มินัล
{ "mcpServers": { "io.github.jghiringhelli-codeseeker": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}claude_desktop_config.json
URL แบบรีโมตหรือคำสั่งบนเครื่องของคุณ? การติดตั้งสองแบบต่างกันอย่างไร
รันบน PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
เปิดจากผู้เผยแพร่
Four-layer hybrid search and knowledge graph for AI coding assistants.
BM25 + vector embeddings + RAPTOR directory summaries + graph expansion — fused into a single MCP tool that gives Claude, Copilot, and Cursor a real understanding of your codebase.
Graph-powered code intelligence with semantic search and knowledge graph for AI assistants
Works with Claude Code, GitHub Copilot (VS Code 1.99+), Cursor, Windsurf, and Claude Desktop.
One command to index; the Claude Code plugin keeps it in sync from there.
Built by PragmaWorks as part of Generative Specification — the discipline for building software with AI that doesn't drift. Two sibling MCP servers compose with this one, each solving a different half of the same problem:
They are independent: install one, or all three.
CodeSeeker solves a retrieval problem, and that problem is shrinking.
If you run a spec-driven loop with a real harness — a specification precise enough that a stateless model derives from it, and quality gates that verify the result against a live system — or an intent-driven one, your assistant is being handed the structure it would otherwise have to go find. A sentinel tool that routes actions instead of sprawling across a dozen endpoints removes another slice of the searching. Frontier models with better navigation and larger effective context remove more. Past a certain point, semantic search over your own code stops being the bottleneck, and this server becomes something you could uninstall without noticing.
That is the intended outcome, not a defect: a method that works should make its own scaffolding unnecessary. Most teams do not start there, though, and until the process is in place CodeSeeker is the cheaper substitute — one command instead of a discipline.
If you would rather have the process than the substitute, start with the field guide: Generative Specification — Field Guide (PDF).
AI assistants are powerful editors, but they navigate code like a tourist:
"find authentication logic" returns every file containing the word "auth"CodeSeeker fixes this. It indexes your codebase once and gives AI assistants a queryable knowledge graph they can use on every turn.
A 4-stage pipeline runs on every query:
Query: "find JWT refresh token logic"
│
▼ Stage 1 — Hybrid retrieval
┌─────────────────────────────────────────────────────┐
│ BM25 (exact symbols, camelCase tokenized) │
│ + │
│ Vector search (384-dim Xenova embeddings) │
│ ↓ │
│ Reciprocal Rank Fusion: score = Σ 1/(60 + rank_i) │
│ Top-30 results, including RAPTOR directory nodes │
└─────────────────────────────────────────────────────┘
│
▼ Stage 2 — RAPTOR cascade (conditional)
┌─────────────────────────────────────────────────────┐
│ IF best directory-summary score ≥ 0.5: │
│ → narrow results to that directory automatically │
│ ELSE: all 30 results pass through unchanged │
│ Effect: "what does auth/ do?" scopes to auth/ │
│ "jwt.ts decode function" bypasses this │
└─────────────────────────────────────────────────────┘
│
▼ Stage 3 — Scoring and deduplication
┌─────────────────────────────────────────────────────┐
│ Dedup: keep highest-score chunk per file │
│ Source files: +0.10 (definition sites matter) │
│ Test files: −0.15 (prevent test dominance) │
│ Symbol boost: +0.20 (query token in filename) │
│ Multi-chunk: up to +0.30 (file has many hits) │
└─────────────────────────────────────────────────────┘
│
▼ Stage 4 — Graph expansion
┌─────────────────────────────────────────────────────┐
│ Top-10 results → follow IMPORTS/CALLS/EXTENDS edges │
│ Structural neighbors scored at source × 0.7 │
│ Avg graph connectivity: 20.8 edges/node │
└─────────────────────────────────────────────────────┘
│
▼
auth/jwt.ts (0.94), auth/refresh.ts (0.89), ...
The knowledge graph is built from AST-parsed imports at index time. It's what powers the graph action, dead-code detection, and graph expansion in every search.
What LSP tools can't do:
What vector-only search misses:
npm install -g codeseeker
claude mcp add codeseeker --scope user -e CODESEEKER_STORAGE_MODE=embedded -- codeseeker serve --mcp
--scope user makes it available in every project you open, not just the current one.
Why global rather than npx -y: on a machine that has never seen the package, npx
downloads it and builds native dependencies before the server can answer, which measured
13.7 seconds to a completed MCP handshake. Clients that give up sooner report that as
a connection failure. A global install answers in 759 ms — the download happens once,
at a moment when you are expecting it to.
Portable, and fine once the package is cached. Expect a slow first start.
{
"mcpServers": {
"codeseeker": {
"command": "npx",
"args": ["-y", "codeseeker", "serve", "--mcp"],
"env": { "CODESEEKER_STORAGE_MODE": "embedded" }
}
}
}
Add this to your MCP config file (see below for per-client locations) and restart your editor.
npm install -g codeseeker
codeseeker install --vscode # or --cursor, --windsurf, --vs
For Claude Code CLI users — adds auto-sync hooks and slash commands:
/plugin install codeseeker@github:jghiringhelli/codeseeker#plugin
Slash commands: /codeseeker:init, /codeseeker:reindex
{
"name": "My Project",
"image": "mcr.microsoft.com/devcontainers/javascript-node:18",
"postCreateCommand": "npm install -g codeseeker && codeseeker install --vscode"
}
Ask your AI assistant: "What CodeSeeker tools do you have?"
You should see a single tool named codeseeker. That is intentional: one tool with an
action routing key keeps per-request token overhead low (ADR-002). The actions are
search, sym, graph, analyze and index.
The MCP config JSON is the same for all clients — only the file location differs:
{
"mcpServers": {
"codeseeker": {
"command": "npx",
"args": ["-y", "codeseeker", "serve", "--mcp"]
}
}
}
npm install -g codeseeker
cd your-project
codeseeker init
codeseeker -c "how does authentication work in this project?"
CodeSeeker exposes one MCP tool, codeseeker. You pick behaviour with action and
fill only the matching nested parameter group:
codeseeker({ action, project, search?|sym?|graph?|analyze?|index? })
Always pass project (the absolute project root) — an MCP server cannot detect your
working directory.
You don't invoke these manually—Claude uses them automatically when searching code or analyzing relationships.
You do not have to index anything first. Searching a project CodeSeeker has never seen starts the index automatically and says so. Indexing runs in the background, so the call returns at once rather than blocking your assistant:
User: "Find the authentication logic"
│
▼
┌────────────────────────────────────────────────────────┐
│ Claude calls codeseeker({action:"search"}) │
│ │ │
│ ▼ │
│ Project indexed? ──No──► starts indexing, returns │
│ │ {status:"indexing_started"} │
│ Yes │
│ ▼ │
│ Return ranked results + confidence │
└────────────────────────────────────────────────────────┘
So the first question you ask a new project answers with "indexing started, retry shortly"
rather than with results. A small project is ready in a couple of seconds; a large one
takes several minutes. Check with index({op:"status"}) rather than guessing.
You can still index deliberately, and it is worth doing for a big repository so the wait happens when you expect it:
codeseeker({ action: "index", index: { op: "init", path: "/abs/path", name: "my-project" } })
codeseeker init # or, from the CLI
name is the only thing an explicit init gives you that the automatic one cannot know —
otherwise the project is registered under its directory name.
The index is a snapshot. Nothing watches your filesystem, so after files change there are three ways it gets back in step:
Without the plugin, an index that drifts will keep returning results; they will simply be
from the code as it was. The stale_index signal is what surfaces that.
18 hand-labelled queries across two real-world codebases:
Each query has one or more mustFind targets (exact file basenames) and optional mustNotFind targets (scope leak check). Queries were run on a real index built from source — real Xenova embeddings, real graph, real RAPTOR L2 nodes — to reflect production conditions.
Metrics: MRR (Mean Reciprocal Rank), P@1 (Precision at 1), R@5 (Recall at 5), F1@3.
BM25 + embedding fusion (RRF)
The workhorse. Handles ~94% of ranking quality on its own. BM25 catches exact symbol names and camelCase tokens; vector embeddings catch semantic similarity when names differ. Fused with Reciprocal Rank Fusion to combine both signals without manual weight tuning.
RAPTOR (hierarchical directory summaries)
Generates per-directory embedding nodes by mean-pooling all file embeddings in a folder. Acts as a post-filter: when a directory summary scores ≥ 0.5 against the query, results are narrowed to that directory's files. Measured contribution: +0.3% MRR on symbol queries. Fires conservatively — only when the directory is an obvious match. Its real value is on abstract queries ("what does the payments module do?") which don't appear in this benchmark; for those queries it prevents broad scattering across the entire codebase.
Knowledge graph (import/dependency edges)
Average connectivity: 20.8 file→file edges per node across both TS and C# codebases. Measured ranking impact: ±0% MRR for 1-hop expansion. The graph doesn't move MRR because the semantic layer already finds the right files — the graph's neighbors are usually already in the top-15. Its value is structural: the analyze dependencies action and explicit graph search type give Claude traversable import chains, inheritance hierarchies, and dependency paths that embeddings alone cannot provide.
Type boost / penalty scoring
Source files get +0.10 score boost; test files get −0.15 penalty; lock files and docs get −0.05 penalty. Without this, integration.test.ts would rank above dag-engine.ts for exact symbol queries because test files import and exercise every symbol in the source. The penalty corrects this without eliminating test files from results.
Monorepo directory exclusion fix
The single highest-impact change in v1.12.0: removing packages/ from the default exclusion list. For pnpm/yarn/lerna monorepos where all source lives under packages/, this exclusion was silently dropping all source files. Effect: 10% → 72% MRR on the Conclave monorepo benchmark.
Reproduce with:
npm run build
node scripts/real-bench.js
Requires C:\workspace\claude\conclave and C:\workspace\ImperialCommander2 to be present locally (or update paths in scripts/real-bench.js).
CodeSeeker analyzes your codebase and extracts patterns:
{
"validation": {
"email": {
"preferred": "z.string().email()",
"usage_count": 12,
"files": ["src/auth.ts", "src/user.ts"]
}
},
"react-patterns": {
"state": {
"preferred": "useState<T>()",
"usage_count": 45
}
}
}
Detected pattern categories:
When Claude writes new code, it follows your existing conventions instead of inventing new ones.
If Claude notices files that shouldn't be indexed (like Unity's Library folder, build outputs, or generated files), it can dynamically exclude them:
// Exclude Unity Library folder and generated files
codeseeker({
action: "index",
project: "/abs/path/to/my-unity-game",
index: {
op: "exclude",
exclude_op: "exclude",
paths: ["Library/**", "Temp/**", "*.generated.cs"],
reason: "Unity build artifacts"
}
})
Exclusions are persisted in .codeseeker/exclusions.json and automatically respected during reindexing.
CodeSeeker helps you maintain a clean codebase by finding duplicate code and detecting dead code.
Ask Claude to find similar code blocks that could be consolidated:
"Find duplicate code in my project"
"Are there any similar functions that could be merged?"
"Show me copy-pasted code that should be refactored"
CodeSeeker uses vector similarity to find semantically similar code—not just exact matches. It detects:
Ask Claude to identify unused code that can be safely removed:
"Find dead code in this project"
"What functions are never called?"
"Show me unused exports"
CodeSeeker analyzes the knowledge graph to find:
Example workflow:
User: "Use CodeSeeker to clean up this project"
Claude: I'll analyze your codebase for cleanup opportunities.
Found 3 duplicate code blocks:
- validateEmail() in auth.ts and user.ts (92% similar)
- formatDate() appears in 4 files with minor variations
- Error handling pattern repeated in api/*.ts
Found 2 dead code files:
- src/utils/legacy-helper.ts (0 imports)
- src/services/unused-service.ts (exported but never imported)
Would you like me to:
1. Consolidate the duplicate validators into a shared utility?
2. Remove the dead code files?
Every language is indexed and searchable — chunking and embeddings do not depend on a parser. What the parser changes is the knowledge graph: how accurately CodeSeeker knows which symbol is a declaration and what depends on what.
Regex extraction is a real limitation, not a smaller version of the same thing. It finds declarations by shape, so it misses anything written unusually and cannot tell a declaration from a call. Import edges stay reliable for TypeScript and JavaScript, where Babel parses them; call edges are heuristic everywhere.
The ratings above are measured, not asserted. scripts/corpus-bench.js indexes seven
real projects and scripts/graph-quality.js reports how much of each resulting graph is
plausibly a real declaration. Java carries no rating of its own because no Java project is
in that corpus; it shares Python's Tree-sitter code path, and that is all we can claim.
Measured on the four RealWorld Conduit implementations — the same application written four times, so a gap between them is language handling and not task difficulty — mean reciprocal rank over 30 labelled queries is TypeScript 83.3%, Python 81.5%, JavaScript 77.1%, C# 70.5%.
If your project is mostly Go, Rust, C++ or Ruby, you can install the Tree-sitter grammar for it:
npm install -g tree-sitter-go # or tree-sitter-rust, tree-sitter-cpp, …
Then ask your assistant, or run:
codeseeker({ action: "index", project: "/abs/path", index: { op: "parsers", list_available: true } })
The response marks each parser with wired: true or wired: false. Only a parser
marked wired: true is consumed by the graph builder — installing one marked false
changes nothing today, and the response says so rather than letting you find out by not
noticing an improvement.
Currently wired: TypeScript, JavaScript, Python, Java, C#. The others are honest false.
Wiring one is a small, self-contained change — a parser implementing
ILanguageParser registered in extensionToParser
(src/mcp/indexing-service.ts) — and contributions are
welcome. scripts/graph-quality.js measures whether a new parser actually improved the
graph, so the improvement is demonstrable rather than assumed.
The plugin installs hooks that automatically update the index:
You don't need to do anything—the plugin handles sync automatically.
codeseeker({action:"index", index:{op:"sync"}})Good fit:
Less useful:
┌──────────────────────────────────────────────────────────┐
│ Claude Code │
│ │ │
│ MCP Protocol │
│ │ │
│ ┌──────────────────────▼──────────────────────────┐ │
│ │ CodeSeeker MCP Server │ │
│ │ ┌─────────────┬─────────────┬────────────────┐ │ │
│ │ │ Vector │ Knowledge │ Coding │ │ │
│ │ │ Search │ Graph │ Standards │ │ │
│ │ │ (SQLite) │ (SQLite) │ (JSON) │ │ │
│ │ └─────────────┴─────────────┴────────────────┘ │ │
│ └─────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────┘
All data stored locally in .codeseeker/. No external services required.
For large teams (100K+ files, shared indexes), server mode supports PostgreSQL + Neo4j. See Storage Documentation.
For the complete technical internals — exact scoring formulas, MCP tool schema, graph edge types, RAPTOR threshold logic, pipeline stages, analysis confidence tiers — see the Technical Architecture Manual.
npx -y codeseeker --versionnode --version (need v18+)First-time indexing of large projects (50K+ files) can take 5+ minutes. Subsequent uses are instant.
Open an issue: GitHub Issues
Claude Code and GitHub Copilot share the same
.vscode/mcp.json— configure once, works for both.
If CodeSeeker is useful to you, consider sponsoring the project.
Free for any use, commercial included — no company-size or revenue limits. Apache-2.0 adds an explicit patent grant over MIT, which is why it is the choice here.
Everything above is the local, single-developer setup: the index lives in .codeseeker/
on your machine and never leaves it.
There is a centralized deployment for teams — one shared index over your organisation's repositories (PostgreSQL + pgvector, Neo4j), so engineers aren't each paying to reindex the same code, and so the graph spans service boundaries instead of stopping at one repo. It also surfaces what the local tool structurally cannot: how code understanding is actually distributed across a codebase and where the knowledge gaps sit.
If that's useful to your team, get in touch: https://pragmaworks.dev
CodeSeeker gives Claude the code understanding that grep and embeddings alone can't provide.
An Apache-2.0 tool behind Generative Specification (GS) — the discipline for building software with AI that doesn't drift: you author a specification precise enough that a stateless AI derives correct code from it, and a harness verifies it against a live system.
Graph-powered code intelligence with semantic search and knowledge graph for AI assistants
{ "mcpServers": { "io.github.jghiringhelli-codeseeker": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}เครื่องมือ
ยังไม่มีรายการเครื่องมือ เซิร์ฟเวอร์แบบรีโมตจะได้รายการหลังการตรวจสอบสถานะครั้งแรก
FAQ
คัดลอกคอนฟิก stdio สำหรับ Cursor, Claude Desktop, Claude Code หรือ VS Code ได้ในหน้านี้ คอนฟิกนี้ใช้การติดตั้งของผู้เผยแพร่ ไม่ใช่พร็อกซีของ PlugMCP
ระเบียนนี้ไม่ได้ระบุตัวแปรสภาพแวดล้อมที่จำเป็น
Codeseeker ทำงานแบบโลคัลผ่าน stdio
ตรวจสอบล่าสุดเมื่อ 2026-11-09 Graph-powered code intelligence with semantic search and knowledge graph for AI assistants