What can you do with this server?

This server provides an authoritative LSL (Linden Scripting Language) reference for AI coding assistants, helping prevent hallucinated functions, wrong signatures, and unsupported syntax. It offers the following tools: * lookup_function(name) — Look up a specific LSL function by name to get its full record: signature, parameters, return type, delay, energy cost, caveats, examples, related functions, and known AI pitfalls. Includes fuzzy matching and did_you_mean suggestions when no exact match is found. * search_functions(query, limit?) — Full-text search across LSL function names and descriptions using natural language or keywords. Returns a ranked summary list. * get_pitfalls(category?, ai_source?) — Retrieve a curated database of known LSL mistakes made by AI coding assistants, filterable by category (e.g. nonexistent_functions, unsupported_syntax) or by AI tool source. * check_code(code) — Scan raw LSL source code for known AI-generated mistakes, including hallucinated function names, unsupported syntax (ternary operators, switch statements), and reserved words used as variable names. Returns line numbers and fix suggestions. * list_events(name?) — List all valid LSL event signatures, or look up a specific event by name to verify it exists and get its parameter details. * get_constants(category?, name?) — Browse or look up LSL constants by category (e.g. permissions, prim_params) or by exact name (e.g. NULL_KEY, PERMISSION_TAKE_CONTROLS).

How do I use lsl-mcp?

1. Click on "Deploy Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@lsl-mcp check this LSL script for common pitfalls" 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.

lsl-mcp

by Treeeeeeeeeeeeee

Overview Schema Source Related Servers Discussions Admin

Python

Local

lsl-mcp

A Model Context Protocol (MCP) server providing accurate LSL (Linden Scripting Language) reference data to AI coding assistants. Grounded in the official Second Life wiki, with a curated database of AI-specific pitfalls.

Motivation

AI coding assistants (Claude Code, Kiro, etc.) frequently produce incorrect LSL — hallucinated function names, unsupported syntax, and wrong signatures. This server gives those tools a live, authoritative reference to query against rather than relying on training data.

Related MCP server: PAELLADOC

Design Goals

Portable LSL only — no Firestorm-specific syntax
AI-aware — pitfalls are categorised and tagged by which tool produced them
Git-tracked — JSON is the source of truth; SQLite is always derivable
Incrementally maintainable — new pitfalls are captured via CLI as they are discovered

Project Structure

lsl-mcp/
├── server.py              # MCP server entrypoint
├── db/
│   ├── schema.sql         # Table definitions
│   └── lsl.db             # SQLite database (gitignored)
├── data/
│   ├── functions/         # One JSON file per LSL function (source of truth)
│   ├── pitfalls.json      # Language pitfalls collection (source of truth)
│   └── constants.json     # LSL constants
├── scripts/
│   ├── scrape_wiki.py     # Populates data/functions/ from the LSL wiki
│   ├── load_db.py         # Imports JSON data into lsl.db
│   └── add_pitfall.py     # CLI tool for adding new pitfalls
├── tools/
│   ├── lookup.py          # lsl_lookup_function, lsl_search
│   ├── pitfalls.py        # lsl_get_pitfalls, lsl_check_code
│   └── reference.py       # lsl_list_events, lsl_constants
└── pyproject.toml

Requirements

Python 3.11+
uv (recommended) or pip

Initial Setup

1. Install dependencies

Using uv (recommended):

uv sync

Using pip:

pip install -e .

For development (includes pytest):

uv sync --dev
# or
pip install -e ".[dev]"

2. Scrape the LSL wiki

Populates data/functions/ with one JSON file per function. Approximately 700 functions; takes several minutes due to polite request throttling.

python scripts/scrape_wiki.py

Test a single function first:

python scripts/scrape_wiki.py --function llListen --dry-run

2. Initialise the database

Creates db/lsl.db from db/schema.sql and loads all JSON data.

python scripts/load_db.py

3. Connect to Claude Code

# using uv (recommended)
claude mcp add --transport stdio lsl -- uv run /path/to/lsl-mcp/server.py

# using python directly
claude mcp add --transport stdio lsl -- python /path/to/lsl-mcp/server.py

Or add a .mcp.json to your project root for project-wide sharing:

{
  "mcpServers": {
    "lsl": {
      "type": "stdio",
      "command": "uv",
      "args": ["run", "/path/to/lsl-mcp/server.py"]
    }
  }
}

Verify the server is connected:

claude mcp list

MCP Tools

Tool	Description
`lookup_function(name)`	Exact or fuzzy match on function name. Returns full record including signature, parameters, caveats, and known AI pitfalls. Falls back to `did_you_mean` on miss.
`search_functions(query, limit?)`	Full-text search across function names and descriptions. Returns ranked summary list.
`get_pitfalls(category?, ai_source?)`	Returns all pitfall entries, optionally filtered by category or source tool.
`check_code(code)`	Scans raw LSL source for known AI-generated mistakes. Returns line numbers and suggestions.
`list_events(name?)`	Returns all valid LSL event signatures, or looks up one by name.
`get_constants(category?, name?)`	Returns constants by category or direct name lookup.

Recommended AI workflow

Before starting an LSL task:
    get_pitfalls()               ← full briefing on known mistakes

While writing LSL:
    lookup_function("llFoo")     ← verify signature before use
    list_events("touch_start")   ← verify event name and parameters
    get_constants("permissions")  ← browse constants by category

Before presenting code to the user:
    check_code(generated_lsl)    ← catch mistakes before they reach the user

Pitfall Categories

Category	Description
`reserved_words`	LSL type names or keywords used as variable identifiers
`nonexistent_functions`	Hallucinated function names with no LSL equivalent
`unsupported_syntax`	Syntax valid in other languages but a compile error in LSL
`scoping`	Permission/access scope issues that silently block operations
`type_coercion`	Implicit casting that LSL does not perform
`state_behavior`	Unexpected behavior around state changes
`performance`	Functions with hidden sleeps or inefficient patterns

Adding a Pitfall

When an AI tool produces incorrect LSL, capture it:

python scripts/add_pitfall.py \
  --category nonexistent_functions \
  --title "llStringReplace does not exist in LSL" \
  --bad "llStringReplace(src, old, new)" \
  --good "llReplaceSubString(src, pattern, replacement, count)" \
  --ai-source kiro \
  --notes "llStringReplace is a plausible-sounding hallucination, likely confused with string replace conventions in other languages."

Then sync the database and commit:

python scripts/load_db.py --only pitfalls
git add data/pitfalls.json
git commit -m "pitfall(func_001): llStringReplace does not exist [kiro]"

Claude-assisted capture

Describe the mistake to Claude in plain language. Claude will format it into the correct schema and emit the exact CLI command to run. You verify, run it, and commit. Claude never writes to the database directly — the CLI is the deterministic protocol.

`--dry-run`

Preview what would be written without touching any files:

python scripts/add_pitfall.py --category unsupported_syntax --title "..." --dry-run

Refreshing Wiki Data

Re-scrape all functions and reload:

python scripts/scrape_wiki.py --overwrite
python scripts/load_db.py --only functions

Re-scrape a single function:

python scripts/scrape_wiki.py --function llReplaceSubString
python scripts/load_db.py --only functions

Database

db/lsl.db is gitignored — it is always fully derivable from the JSON sources via load_db.py. Only data/ and db/schema.sql are version-controlled.

To rebuild from scratch:

python scripts/load_db.py --reset

Pitfall ID Format

IDs are assigned automatically by add_pitfall.py based on category:

Category	Prefix	Example
`reserved_words`	`lang`	`lang_001`
`nonexistent_functions`	`func`	`func_001`
`unsupported_syntax`	`syn`	`syn_001`
`scoping`	`scope`	`scope_001`
`type_coercion`	`type`	`type_001`
`state_behavior`	`state`	`state_001`
`performance`	`perf`	`perf_001`

Known Pitfalls

ID	Category	Issue	Source
`type_001`	`type_coercion`	Boolean NOT `!` does not work on strings	claude-code
`state_001`	`state_behavior`	Cached owner key not updated after ownership transfer	claude-code
`lang_001`	`reserved_words`	`key` used as variable name	kiro
`func_001`	`nonexistent_functions`	`llStringReplace` does not exist	kiro
`syn_001`	`unsupported_syntax`	Ternary operator not supported	both
`syn_002`	`unsupported_syntax`	Switch statements not supported	both
`scope_001`	`scoping`	Transfer-only permissions silently prevent script edits	kiro
`perf_001`	`state_behavior`	`llSetPrimitiveParams` has forced 0.2s sleep; use `llSetLinkPrimitiveParamsFast`	both

Testing

# Run all tests
python3 -m pytest

# Verbose output
python3 -m pytest -v

# Single module
python3 -m pytest tests/test_pitfalls.py

Tests use an in-memory SQLite fixture database built from db/schema.sql — no real lsl.db or network access required. The conftest.py fixture patches DB_PATH in all tool modules automatically.

Test file	Coverage
`test_lookup.py`	`lsl_lookup_function`, `lsl_search`
`test_pitfalls.py`	`lsl_get_pitfalls`, `lsl_check_code`
`test_reference.py`	`lsl_list_events`, `lsl_constants`
`test_add_pitfall.py`	ID generation logic in `add_pitfall.py`

gitignore

db/lsl.db
data/functions/
data/scrape_errors.json
__pycache__/
*.pyc
.venv/

data/functions/ is gitignored because it is fully regenerable from the wiki. Only data/pitfalls.json and data/constants.json are committed — these contain hand-curated data that cannot be scraped.

Available Tools

6 tools

check_codeA

Scan an LSL code snippet for known AI-generated pitfalls.

Checks for nonexistent function calls, unsupported syntax (ternary operators, switch statements), reserved words used as variable names, and other patterns from the pitfalls database.

Call this on any LSL you generate before presenting it to the user. Returns line numbers and suggestions for each issue found.

Args: code: Raw LSL source code as a string.

Inputs: code (required).

get_constantsA

Return LSL constants, optionally filtered by category or name.

Call with no arguments to see available categories and total count. Use category to browse a group (e.g. "permissions", "prim_params"). Use name for a direct lookup (e.g. "NULL_KEY", "PERMISSION_TAKE_CONTROLS").

Args: category: Optional category filter. See response for valid categories. name: Optional exact constant name. Takes precedence over category.

Inputs: category, name.

get_pitfallsA

Return known LSL pitfalls for AI coding assistants.

Call with no arguments for a full briefing before starting an LSL task. Filter by category or by which AI tool produced the mistake.

Inputs: category, ai_source.

list_eventsA

Return valid LSL event signatures.

Call with no arguments to get all events and verify an event name exists. Call with a name to get the full signature and parameter details.

Args: name: Optional event name, e.g. "listen" or "touch_start". Omit to return all events.

Inputs: name.

lookup_functionA

Look up an LSL function by name.

Returns the full function record: signature, parameters, return type, delay, energy cost, caveats, examples, related functions, and any known AI-specific pitfalls associated with this function.

Falls back to fuzzy matching if the exact name is not found, and returns a 'did_you_mean' list when no match exists at all — helping catch hallucinated function names.

Args: name: Function name, e.g. "llListen" or "llReplaceSubString".

Inputs: name (required).

search_functionsA

Full-text search across LSL function names and descriptions.

Use when you know roughly what a function does but not its exact name. Returns a ranked summary list — call lookup_function for the full record.

Args: query: Keywords or natural language, e.g. "listen channel message" or "set prim texture face". limit: Maximum results to return (default 10, max 25).

Inputs: query (required), limit.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

Deploy Server

Claude Code, inside the Godot 4 editorFlockbay puts an AI agent in the real Godot 4 editor: it writes GDScript, runs your game and fixes what breaks. Free, or bring your own Claude Code or Codex.flockbay.comAd

TDQS

A4.6/5.0

Scored 2026-05-16 01:15 across 6 tools

Disambiguation5/5

Each tool targets a distinct aspect of LSL development: code checking, constants, pitfalls database, events, function lookup by name, and full-text search. No two tools have overlapping purposes, and descriptions clearly differentiate them.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern in lowercase snake_case (e.g., check_code, lookup_function). The naming is predictable and easy to understand.

Tool Count5/5

With 6 tools, the server is well-scoped for assisting with LSL code. Each tool provides essential functionality without unnecessary bloat. The count feels appropriate for the domain.

Completeness5/5

The tool set covers the core needs for LSL development: checking for pitfalls, exploring constants, events, and functions. Missing features like syntax guidance are mitigated by detailed returns from lookup_function and search_functions.

Maintenance

ActivityInactive

ResponsivenessLimited data

Resources

Related MCP Connectors

hithereiamaliff-mcp-nextcloud
ai.smithery
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
Cloud Storage File Systems Workplace & Productivity
239
https://server.smithery.ai/@hithereiamaliff/mcp-nextcloud/mcp
ChiR24-unreal_mcp_server
ai.smithery
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to control Unreal E…
Developer Tools App Automation Software Architecture
21,032
https://server.smithery.ai/@ChiR24/unreal_mcp_server/mcp
UnifAPI
com.unifapi
Hosted MCP server for live public-data APIs and Skills for AI agents.
Open Data Search AI & Machine Learning
2
https://mcp.unifapi.com
slima
io.github.slima-ai
MCP Server for Slima - AI Writing IDE for Novel Authors with AI Beta Reader.
Developer Tools
21
https://mcp.slima.ai/mcp

Related MCP Servers

MCP Server Example
Developer Tools Code Execution Autonomous Agents
alejandro-ao
A
license
B
quality
D
maintenance
An educational implementation of a Model Context Protocol server that demonstrates how to build a functional MCP server for integrating with various LLM clients like Claude Desktop.
Updated 2 years ago (2025-04-29 10:58 UTC)
1
161
MIT
PAELLADOC
Developer Tools Documentation Access Autonomous Agents
jlcases
A
license
Not graded
quality
F
maintenance
A Model Context Protocol (MCP) server that implements AI-First Development framework principles, allowing LLMs to interact with context-first documentation tools and workflows for preserving knowledge and intent alongside code.
Updated 6 months ago (2026-05-19 16:14 UTC)
336
AGPL 3.0
RL-MCP
RAG Systems Agent Orchestration Databases
rlefko
F
license
Not graded
quality
D
maintenance
A Model Context Protocol server that provides AI models with structured access to external data and services, acting as a bridge between AI assistants and applications, databases, and APIs in a standardized, secure way.
Updated a year ago (2025-05-31 18:35 UTC)
2
-
Gravitas-Core-MCP
Coding Agents Autonomous Agents
ahmed-coding
A
license
B
quality
D
maintenance
Production-grade, autonomous Model Context Protocol (MCP) server that elevates AI models from stateless code generators into persistent, self-verifying software engineers.
Updated 9 months ago (2026-02-11 05:04 UTC)
21
1
MIT

lsl-mcp

Motivation

Design Goals

Project Structure

Requirements

Initial Setup

1. Install dependencies

2. Scrape the LSL wiki

2. Initialise the database

3. Connect to Claude Code

MCP Tools

Recommended AI workflow

Pitfall Categories

Adding a Pitfall

Claude-assisted capture

--dry-run

Refreshing Wiki Data

Database

Pitfall ID Format

Known Pitfalls

Testing

gitignore

Available Tools

Tool Schema Changelog

TDQS

Maintenance

Resources

Related MCP Connectors

Related MCP Servers

MCP Server Example

PAELLADOC

RL-MCP

Gravitas-Core-MCP

`--dry-run`