Skip to main content
Glama

FluffOS MCP Server

Real driver validation for LPC development - An MCP server that wraps FluffOS CLI tools to provide actual driver-level validation and debugging.

This MCP server exposes FluffOS's powerful CLI utilities (symbol and lpcc) to AI assistants, enabling them to validate LPC code against the actual driver and examine compiled bytecode.

What This Enables

AI assistants can now:

  • Validate LPC files using the actual FluffOS driver (not just syntax checking)

  • Catch runtime compilation issues that static analysis misses

  • Examine compiled bytecode to debug performance or behavior issues

  • Understand how LPC code actually compiles

Related MCP server: LPC MCP Server

Tools

  • fluffos_validate: Validate an LPC file using FluffOS's symbol tool

  • fluffos_disassemble: Disassemble LPC to bytecode using lpcc

  • fluffos_doc_lookup: Search FluffOS documentation for efuns, applies, concepts, etc.

  • fluffos_eval: Evaluate LPC statements against the live driver using lpcshell (opt-in)

fluffos_validate, fluffos_disassemble, and fluffos_doc_lookup are read-only and idempotent — they never modify files, drivers, or running MUDs, and are safe for agents to auto-invoke.

fluffos_eval is not read-only: it boots the full runtime and executes the LPC you give it, so it can have side effects (writing files, mutating daemon/database state, firing events). It is registered only when FLUFFOS_ENABLE_EVAL is set, and should not be auto-invoked on untrusted input.

When to use which tool

I want to…

Use

Check whether a file compiles against the driver

fluffos_validate

See the bytecode a function compiles to

fluffos_disassemble

Investigate why a pattern is slow

fluffos_disassemble

Look up an efun signature or apply semantics

fluffos_doc_lookup

Find out if an efun exists in this driver build

fluffos_validate on a file that calls it

Pre-commit / pre-deploy sanity check

fluffos_validate

See the actual runtime value/behaviour of an expression

fluffos_eval

Reproduce a runtime error interactively

fluffos_eval

fluffos_doc_lookup is only registered when the server is started with FLUFFOS_DOCS_DIR set. fluffos_eval is only registered when FLUFFOS_ENABLE_EVAL is set.

Prerequisites

1. FluffOS Installation

You need FluffOS installed with the CLI tools available. The following binaries should exist:

  • symbol - For validating LPC files

  • lpcc - For disassembling to bytecode

  • lpcshell - (Optional) For fluffos_eval; required only when FLUFFOS_ENABLE_EVAL is set

2. Node.js

Node.js 16+ required:

node --version  # Should be v16.0.0 or higher

Installation

You can install the server via npm:

npm install -g @gesslar/fluffos-mcp

Or clone and install locally:

git clone https://github.com/gesslar/fluffos-mcp.git
cd fluffos-mcp
npm install

Configuration

The server requires these environment variables:

  • FLUFFOS_BIN_DIR - Directory containing FluffOS binaries (symbol, lpcc, and optionally lpcshell)

  • MUD_RUNTIME_CONFIG_FILE - Path to your FluffOS config file (e.g., /mud/lib/etc/config.test)

  • FLUFFOS_DOCS_DIR - (Optional) Directory containing FluffOS documentation for doc lookup

  • FLUFFOS_ENABLE_EVAL - (Optional) Set to true (or 1/yes/on, case-insensitive) to register fluffos_eval, which executes live LPC via lpcshell. Off by default — any other value, including false/0 or leaving it unset, keeps the tool disabled because it is not read-only.

  • FLUFFOS_EVAL_TIMEOUT_MS - (Optional) Wall-clock cap in milliseconds for a single fluffos_eval run before the lpcshell child is killed. Defaults to 30000 (30s); ignored unless fluffos_eval is enabled.

  • FLUFFOS_EVAL_MAX_BYTES - (Optional) Maximum size in bytes of an fluffos_eval code payload; larger requests are rejected before anything is written to disk. Defaults to 10485760 (10 MiB); ignored unless fluffos_eval is enabled.

  • FLUFFOS_EVAL_MAX_CONCURRENT - (Optional) Maximum number of fluffos_eval runs allowed in flight at once; further requests are rejected until a slot frees. Each run spawns a full lpcshell driver boot, so this bounds both temp-storage use and driver load. Defaults to 4; ignored unless fluffos_eval is enabled.

Setup for Different AI Tools

Warp (Terminal)

Add to your Warp MCP configuration:

Location: Settings → AI → Model Context Protocol

If installed via npm:

{
  "fluffos": {
    "command": "npx",
    "args": ["@gesslar/fluffos-mcp"],
    "env": {
      "FLUFFOS_BIN_DIR": "/path/to/fluffos/bin",
      "MUD_RUNTIME_CONFIG_FILE": "/mud/lib/etc/config.test",
      "FLUFFOS_DOCS_DIR": "/path/to/fluffos/docs"
    }
  }
}

If cloned locally:

{
  "fluffos": {
    "command": "node",
    "args": ["/absolute/path/to/fluffos-mcp/index.js"],
    "env": {
      "FLUFFOS_BIN_DIR": "/path/to/fluffos/bin",
      "MUD_RUNTIME_CONFIG_FILE": "/mud/lib/etc/config.test",
      "FLUFFOS_DOCS_DIR": "/path/to/fluffos/docs"
    }
  }
}

Important: Use absolute paths!

Restart Warp after adding the configuration.

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or equivalent:

If installed via npm:

{
  "mcpServers": {
    "fluffos": {
      "command": "npx",
      "args": ["@gesslar/fluffos-mcp"],
      "env": {
        "FLUFFOS_BIN_DIR": "/path/to/fluffos/bin",
        "MUD_RUNTIME_CONFIG_FILE": "/mud/lib/etc/config.test",
        "FLUFFOS_DOCS_DIR": "/path/to/fluffos/docs"
      }
    }
  }
}

If cloned locally:

{
  "mcpServers": {
    "fluffos": {
      "command": "node",
      "args": ["/absolute/path/to/fluffos-mcp/index.js"],
      "env": {
        "FLUFFOS_BIN_DIR": "/path/to/fluffos/bin",
        "MUD_RUNTIME_CONFIG_FILE": "/mud/lib/etc/config.test",
        "FLUFFOS_DOCS_DIR": "/path/to/fluffos/docs"
      }
    }
  }
}

Restart Claude Desktop after configuration.

Enabling live LPC eval (optional)

The examples above register only the three read-only tools. To also expose fluffos_eval — which boots the driver and executes LPC, so it can have side effects — add FLUFFOS_ENABLE_EVAL to the env block alongside the others:

"env": {
  "FLUFFOS_BIN_DIR": "/path/to/fluffos/bin",
  "MUD_RUNTIME_CONFIG_FILE": "/mud/lib/etc/config.test",
  "FLUFFOS_ENABLE_EVAL": "true"
}

Accepted "on" values are true, 1, yes, or on (case-insensitive). Any other value — or omitting the variable entirely — leaves the tool disabled. The lpcshell binary must exist in FLUFFOS_BIN_DIR for this to work.

Usage Examples

Once configured, you can ask your AI assistant:

"Validate this LPC file with the actual driver" → AI uses fluffos_validate to run symbol

"Show me the bytecode for this function" → AI uses fluffos_disassemble to run lpcc

"Why is this code slow?" → AI examines the disassembly to identify inefficient patterns

"What's the syntax for call_out?" → AI uses fluffos_doc_lookup to search documentation

"How do I use mappings?" → AI searches docs for mapping-related documentation

How It Works

AI Assistant
    ↓ (natural language)
  MCP Protocol
    ↓ (tool calls: fluffos_validate, fluffos_disassemble)
  This Server
    ↓ (spawns: symbol, lpcc)
  FluffOS CLI Tools
    ↓ (validates/compiles with actual driver)
  Your LPC Code
  1. AI assistant sends MCP tool requests

  2. Server spawns appropriate FluffOS CLI tool

  3. CLI tool validates/disassembles using the driver

  4. Server returns results to AI

  5. AI understands your code at the driver level and can reference FluffOS documentation to explain how functions work!

Implementation Details

Architecture

The server is built using the Model Context Protocol SDK and follows a class-based architecture:

  • FluffOSMCPServer class: Main server implementation

  • MCP SDK Server: Handles protocol communication via stdio

  • Child process spawning: Executes FluffOS CLI tools

  • Path normalization: Converts absolute paths to mudlib-relative paths

Path Handling

The server intelligently handles file paths:

  1. Parses mudlib directory from your FluffOS config file

  2. Normalizes absolute paths to mudlib-relative paths

  3. Passes normalized paths to FluffOS tools (which expect relative paths)

Example: /mud/ox/lib/std/object.cstd/object.c

Tool Implementation

fluffos_validate:

  • Spawns symbol <config> <file> from the config directory

  • Captures stdout/stderr

  • Returns success/failure with compilation errors

  • Exit code 0 = validation passed

fluffos_disassemble:

  • Spawns lpcc <config> <file> from the config directory

  • Returns complete bytecode disassembly

  • Includes function tables, strings, and instruction-level detail

fluffos_doc_lookup (optional):

  • Runs scripts/search_docs.sh helper script

  • Uses grep to search markdown files

  • Only available if FLUFFOS_DOCS_DIR is set

fluffos_eval (optional):

  • Writes the submitted LPC statements to a temporary script file, then spawns lpcshell <config> <tmpfile>

  • The temp file is a plain OS file (read by lpcshell directly, not through the driver's file system) and so does not need to live inside the mudlib jail — only the LPC statements execute in-jail

  • Boots the full runtime and executes the code; captures stdout/stderr and exit code, then deletes the temp file

  • Only available if FLUFFOS_ENABLE_EVAL is set

Error Handling

  • Validates required environment variables on startup

  • Returns structured error responses via MCP

  • Gracefully handles missing config or tool execution failures

  • Non-zero exit codes are reported but don't crash the server

Complementary Tools

This server works great alongside:

  • lpc-mcp - Language server integration for code intelligence

  • VS Code with jlchmura's LPC extension - IDE support

Use them together for the complete LPC development experience!

Contributing

PRs welcome! This is a simple wrapper that can be extended with more FluffOS tools.

Credits

License

@gesslar/fluffos-mcp is released under the 0BSD.

This package includes or depends on third-party components under their own licenses:

Available Tools

2 tools
fluffos_disassembleB

Disassemble an LPC file to show compiled bytecode using lpcc. Useful for debugging and understanding how code compiles.

ParametersJSON Schema
NameRequiredDescriptionDefault
fileYesAbsolute path to the LPC file to disassemble

TDQS

B3.3/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the full burden. It mentions the tool is 'useful for debugging and understanding how code compiles,' which gives some behavioral context about its purpose. However, it doesn't disclose important behavioral traits like whether this is a read-only operation, what permissions are needed, whether it modifies the file, error handling, or output format details.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is appropriately sized and front-loaded. The first sentence clearly states the core functionality, and the second sentence adds valuable context about when to use it. Both sentences earn their place with zero waste.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool has 1 parameter with 100% schema coverage and no output schema, the description provides adequate but minimal context. It explains what the tool does and its use cases, but for a tool with no annotations and no output schema, it should ideally provide more behavioral details about what the disassembly output looks like or any limitations.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema description coverage is 100%, with the single parameter 'file' well-documented in the schema. The description doesn't add any parameter-specific information beyond what the schema provides. With high schema coverage, the baseline score of 3 is appropriate as the description doesn't compensate but doesn't need to.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: 'Disassemble an LPC file to show compiled bytecode using lpcc.' It specifies the verb (disassemble), resource (LPC file), and method (using lpcc). However, it doesn't explicitly differentiate from the sibling tool 'fluffos_validate', which appears to be a different operation.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides implied usage context: 'Useful for debugging and understanding how code compiles.' This suggests when to use the tool, but doesn't explicitly state when not to use it or mention alternatives. No comparison to the sibling tool is provided.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

fluffos_validateA

Validate an LPC file using the FluffOS driver's symbol tool. Returns success/failure and any compilation errors.

ParametersJSON Schema
NameRequiredDescriptionDefault
fileYesAbsolute path to the LPC file to validate

TDQS

A3.5/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden of behavioral disclosure. It mentions the validation outcome ('success/failure and any compilation errors'), which is useful context beyond basic functionality. However, it doesn't address important behavioral aspects like error handling, performance characteristics, or whether this operation has side effects.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is perfectly concise with two sentences that each earn their place: the first states the core functionality, and the second describes the return value. There's zero waste or redundancy, and it's appropriately front-loaded with the main purpose.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a single-parameter validation tool with no annotations and no output schema, the description provides adequate but minimal information. It covers what the tool does and what it returns, but doesn't address potential complexities like what constitutes validation success versus failure, error message formats, or limitations of the validation process.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already fully documents the single 'file' parameter. The description doesn't add any additional meaning about the parameter beyond what the schema provides, such as file format requirements or validation scope limitations. The baseline of 3 is appropriate when the schema does all the parameter documentation work.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the specific action ('validate an LPC file') and resource ('using the FluffOS driver's symbol tool'), distinguishing it from the sibling tool 'fluffos_disassemble' which implies a different operation. It provides a complete verb+resource+scope statement.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance is provided on when to use this tool versus alternatives. While the sibling tool name suggests a different function, the description doesn't explicitly state when to choose validation over disassembly or other potential options. There's no mention of prerequisites or context for usage.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.

  1. 2 tool updatesv1.0.0
    • First observedfluffos_disassemble
    • First observedfluffos_validate

TDQS

B3.4/5.0
Disambiguation5/5

The two tools have clearly distinct purposes: disassemble focuses on showing compiled bytecode for debugging, while validate checks for compilation errors and success/failure. There is no overlap or ambiguity between these functions.

Naming Consistency5/5

Both tools follow a consistent 'fluffos_verb' naming pattern (fluffos_disassemble and fluffos_validate), using the same prefix and verb-based structure. This makes them predictable and easy to identify.

Tool Count2/5

With only 2 tools, the server feels thin for a FluffOS/LPC development environment. While disassembly and validation are useful, typical development workflows would benefit from additional tools like code execution, file management, or debugging support, making this set under-scoped.

Completeness2/5

The server covers only disassembly and validation, leaving significant gaps for a FluffOS development tool. Missing are core operations like running LPC code, editing files, managing projects, or advanced debugging features, which limits agents' ability to handle comprehensive development tasks.

Maintenance

ActivityActive
ResponsivenessNo issues

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

Related MCP Servers

  • F
    license
    Not graded
    quality
    Not graded
    maintenance
    Enables AI agents to compile, execute, and check Almide code for syntax or compilation errors. It provides tools for generating ASTs and accessing language grammar resources.
    -
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables AI assistants to understand and query LPC codebases through natural language, using a language server for hover info, definitions, references, and diagnostics.
    13
    2
    BSD Zero Clause

Latest Blog Posts

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/gesslar/fluffos-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server