ministic-fishstick
Integrates with GitHub Copilot Agents in VS Code to expose semantic code search tools to AI assistants.
Supports Ollama as a local embedding provider for semantic code search without external API dependencies.
Uses OpenAI's text embedding models to enable semantic code indexing and vector search.
Provides a zero-docker local vector store using SQLite for semantic code indexing and search.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@ministic-fishstickfind where vector embeddings are stored"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
ministic-fishstick
Minimal, high-performance Model Context Protocol (MCP) server for semantic code indexing and vector search powered by Bun and TypeScript.
ministic-fishstick extracts code-indexing capabilities into a standalone MCP server that can be used directly with AI CLI agents (OpenCode, Claude Desktop, Cursor) as well as GitHub Copilot Agents in VS Code.
Features
🚀 Bun Native & Fast: Built for Bun with zero-lock overhead and fast execution.
💾 Zero-Docker SQLite Vector Search: Native
bun:sqlitevector storage with Float32Array cosine similarity — runs locally without needing Docker or external database services.🔌 Qdrant Support: Optional fallback or upgrade to a remote or local Qdrant vector database (
@qdrant/js-client-rest).🌳 Tree-Sitter AST Parsing: Accurate semantic code block extraction for 25+ programming languages using
web-tree-sitterandtree-sitter-wasms.🛡️ Smart Ignore Rules (
.fishignore+.gitignore): Respects both workspace.gitignoreand.fishignorepatterns to exclude sensitive files or build output.⚙️ Tiered Configuration System: Resolves configuration seamlessly across runtime MCP tool overrides, workspace
.fishstick.json, global~/.config/fishstick/fishstick.json,.envfiles, and defaults.🤖 VS Code & Copilot Agent Ready: Includes a
vscode-extension/wrapper exposing native VS Code MCP Server contributions (contributes.mcpServers) and Copilot LM tools.
Related MCP server: Satori
Default Behavior
When started without extra configuration:
Embedding Provider: Defaults to OpenAI using
text-embedding-3-small(requiresOPENAI_API_KEY).Vector Store: Defaults to local zero-docker
sqlitestorage located at<workspace>/.fishstick/vectors.sqlite.Local Cache: Incremental file scan hashes stored in
<workspace>/.fishstick/cache.sqlite.Target Directory: Indexes the current working directory (
process.cwd()).
Getting Started
Prerequisites
Bun (v1.1+): Install via
curl -fsSL https://bun.sh/install | bashAn OpenAI API Key (or an alternative supported provider like Ollama, Gemini, Mistral, Bedrock, etc.)
Installation & Local Setup
# Clone the repository
git clone https://github.com/your-org/ministic-fishstick.git
cd ministic-fishstick
# Install dependencies
bun install
# Run tests
bun test
# Type-check
bun check-typesUsage
1. Running as an MCP Server (Stdio)
Start the stdio MCP server directly using Bun:
OPENAI_API_KEY="sk-..." bun run src/index.ts2. Configuring in OpenCode / Claude Desktop / Cursor
Add ministic-fishstick to your MCP client configuration (e.g. opencode.json or claude_desktop_config.json):
{
"mcpServers": {
"fishstick": {
"command": "bun",
"args": ["run", "/path/to/ministic-fishstick/src/index.ts"],
"env": {
"OPENAI_API_KEY": "sk-...",
"VECTOR_STORE_PROVIDER": "sqlite"
}
}
}
}3. VS Code Copilot Agent Integration
For VS Code users, the included vscode-extension/ folder provides an extension wrapper:
Open
vscode-extension/in VS Code or install the compiled extension package.The extension automatically registers
fishstickin VS Code's native MCP server catalog and exposes thefishstick_search_codetool to GitHub Copilot chat participants and agents.
Available MCP Tools
The server exposes five core MCP tools:
MCP Tool Name | Description |
| Perform semantic vector search over the indexed codebase. Accepts |
| Trigger background directory scan and file watcher ( |
| Retrieve current indexing state ( |
| Clear vector database tables and local cache files for a workspace. |
| Dynamically update embedding provider, model ID, search minScore, maxResults, or vector store at runtime. |
Configuration Hierarchy
Configuration is resolved automatically in the following order of precedence (highest to lowest):
┌─────────────────────────────────────────────────────────┐
│ Priority 1: Dynamic MCP Tool Calls (`code_index_config`)│
├─────────────────────────────────────────────────────────┤
│ Priority 2: Workspace Config (`.fishstick.json`) │
├─────────────────────────────────────────────────────────┤
│ Priority 3: Global User Config │
│ (`~/.config/fishstick/fishstick.json`) │
├─────────────────────────────────────────────────────────┤
│ Priority 4: Environment Variables (`.env`) / Defaults │
└─────────────────────────────────────────────────────────┘Example .fishstick.json or ~/.config/fishstick/fishstick.json
{
"enabled": true,
"vectorStore": {
"provider": "sqlite",
"qdrantUrl": "http://localhost:6333"
},
"embedder": {
"provider": "openai",
"modelId": "text-embedding-3-small",
"apiKey": "sk-..."
},
"search": {
"minScore": 0.3,
"maxResults": 20
}
}Environment Variables (.env)
Variable | Description | Default |
| OpenAI API Key for default embedder | — |
|
|
|
| Qdrant database server URL |
|
| Embedder provider ( |
|
| Model identifier |
|
| Similarity score cutoff (0.0 to 1.0) |
|
| Max results returned by search |
|
Ignore Rules (.fishignore + .gitignore)
ministic-fishstick uses FishIgnoreController to filter files before parsing and indexing:
.gitignore: Automatically respected if present in the workspace..fishignore: Custom ignore file for indexing rules (uses standard.gitignoreglob syntax).Auto-ignored:
.git,.fishignore,.fishstick.json,node_modules/, binary files, and vector database caches are always excluded.
License
Apache 2.0 © 2026 Israel Flores-Arbolay.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Alicense-qualityDmaintenanceLocal MCP server for indexing personal knowledge into SQLite with hybrid search, chunk-level citations, memory tools, and agent orchestration.4MIT
- Alicense-qualityBmaintenanceAgent-safe code retrieval MCP server that indexes repositories and provides semantic search, file navigation, call graph analysis, and bounded file reading tools for coding agents.3,977,9623AGPL 3.0
- Alicense-qualityCmaintenanceAn MCP server that indexes your codebase using tree-sitter AST parsing and gives AI tools instant access to structural intelligence like dependency graphs, call trees, and dead code detection from a local SQLite database.MIT
- FlicenseBqualityBmaintenanceA local MCP server that indexes codebases using Tree-sitter AST parsing and LanceDB, enabling AI models to search across repositories with sub-second latency.5
Related MCP Connectors
Agent-native MCP server over the public saagarpatel.dev corpus. Read-only, stateless.
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/israelflores8789/ministic-fishstick'
If you have feedback or need assistance with the MCP directory API, please join our Discord server