Laisky is a remote MCP server that gives an AI agent 20 tools, among them extract_key_info, file_delete, file_list. It answers at https://mcp.laisky.com over streamable HTTP, and it listed its tools to an anonymous client, so no sign-in was needed for that. Our probes reached it 4 of 4 times since 23 September 2026; the last check, on 6 November 2026, succeeded.
| Tool | What it does |
|---|---|
extract_key_info | Extract the most relevant context chunks for a query from provided materials using RAG (retrieval-augmented generation). Splits text into chunks, embeds them, and returns the top-K most relevant passages. Inputs: materials, query, top_k |
file_delete | Delete one file using its required expected_version from file_read or file_stat. Recursive directory mutations are not supported by file-version tokens; list and delete individual files with their own versions. The root cannot be deleted. Inputs: expected_version, path, plugin, project, recursive |
file_list | List files and directories under a path. Use this to browse, explore, or enumerate directory contents with configurable depth. Inputs: depth, limit, path, plugin, project |
file_list_versions | List immutable file history, newest ID first. Results are paginated; pass next_cursor as before_id. IDs are decimal strings, never live file versions. Inputs: before_id, limit, path, project |
file_read | Read UTF-8 content and its version from one snapshot. Pin subsequent ranges with expected_version to reject mixed-generation reads. Byte ranges must align with UTF-8 characters. Inputs: expected_version, length, offset, path, plugin, project |
file_read_version | Read one immutable history snapshot. Legacy non-UTF-8 bytes return base64 and must not be treated as UTF-8 text. Inputs: history_id, path, project |
file_rename | Rename one file using its required source expected_version. Non-overwriting moves require an absent destination. Overwriting moves also require expected_destination_version or destination_must_not_exist=true. Directory moves are not support Inputs: destination_must_not_exist, expected_destination_version, expected_version, from_path, overwrite, plugin |
file_restore_version | Restore immutable history as a new live revision. Supply the ORIGINAL current-file expected_version, or create_only=true when absent. A conflict has no write effect; re-read and review, never silently update the token. Uses the selected pro Inputs: create_only, expected_version, history_id, path, plugin, project |
file_search | Search file content using hybrid retrieval (semantic + keyword). Use this to find text, code, or patterns within files, similar to grep or full-text search. Each hit returns the matched chunk plus a concise file-level summary in file_summar Inputs: limit, path_prefix, plugin, project, query |
file_stat | Return metadata and the live version for a file. Directories have no file version. For editing, use the version returned with file_read content, not a later independent stat. Inputs: path, plugin, project |
file_write | Write UTF-8 file content and return its committed version. Every call, including APPEND, MUST supply expected_version from a prior read or create_only=true for a new file. Missing conditions fail; stale versions require re-reading and recom Inputs: content, content_encoding, create_only, expected_version, mode, offset |
find_tool | Search for available MCP tools by query. Supports multiple search modes: regex (pattern matching on tool names/descriptions), bm25 (keyword relevance ranking), embedding (semantic similarity via embeddings), or auto (automatically selects t Inputs: mode, query, return_references_only |
get_user_request | During the execution of a task, get the latest user instructions to adjust agent's work objectives or processes. The response may include text, inline images, and resource links when the user attached images. Inputs: task_id |
mcp_pipe | Execute a pipeline of MCP tools (sequential, parallel, nested) with output passing. Inputs: spec |
memory_after_turn | Persist turn artifacts and update memory tiers after model response. Inputs: conversation_items, current_input_count, current_input_start, input_items, output_items, plugin |
memory_before_turn | Prepare model input with recalled memory context for the current turn. Inputs: base_instructions, conversation_items, current_input, current_input_count, current_input_start, current_input_text |
memory_list_dir_with_abstract | List memory directories and include abstract/overview metadata. Inputs: depth, limit, path, plugin, project, session_id |
memory_run_maintenance | Run compaction, retention sweep, and summary refresh for one memory session. Inputs: plugin, project, session_id |
web_fetch | Fetch and render dynamic web content by URL. Downloads a web page and returns its content as HTML or Markdown. Use this to retrieve, read, or scrape a specific webpage. Inputs: output_markdown, url |
web_search | Search the public web using search engines (Google, Bing) and return a structured result set. Use this to find information, articles, documentation, or answers from the internet. Inputs: query |
Every block below points at https://mcp.laisky.com. Pick your client and copy.
claude mcp add --transport http laisky https://mcp.laisky.com
{
"mcpServers": {
"laisky": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.laisky.com"]
}
}
}{
"mcpServers": {
"laisky": {
"url": "https://mcp.laisky.com"
}
}
}{
"servers": {
"laisky": {
"type": "http",
"url": "https://mcp.laisky.com"
}
}
}{
"context_servers": {
"laisky": {
"url": "https://mcp.laisky.com"
}
}
}{
"mcpServers": {
"laisky": {
"serverUrl": "https://mcp.laisky.com"
}
}
}{
"mcpServers": {
"laisky": {
"url": "https://mcp.laisky.com"
}
}
}name: Laisky
version: 0.0.1
schema: v1
mcpServers:
- name: Laisky
type: streamable-http
url: https://mcp.laisky.com{
"mcpServers": {
"laisky": {
"httpUrl": "https://mcp.laisky.com"
}
}
}{
"mcpServers": {
"laisky": {
"type": "http",
"url": "https://mcp.laisky.com"
}
}
}{
"mcpServers": {
"laisky": {
"url": "https://mcp.laisky.com"
}
}
}goose://extension?url=https%3A%2F%2Fmcp.laisky.com&type=streamable_http&id=laisky&name=Laisky
It answered our last check on 6 November 2026 in 606 ms. Over 4 checks since 23 September 2026 it answered 4. Status here comes from Neuronto's own probes, not from the publisher.
Yes. It is published by laisky.com and served from mcp.laisky.com, the vendor's own domain: https://mcp.laisky.com.
20, as returned by the server's own tools/list on 9 November 2026: extract_key_info, file_delete, file_list, file_list_versions, file_read, file_read_version, file_rename, file_restore_version, and 12 more.
It answered our last check on 6 November 2026 in 606 ms. Over 4 checks since 23 September 2026 it answered 4.
Put the server under mcpServers in ~/.cursor/mcp.json with the URL https://mcp.laisky.com. The exact JSON is above, with the same setup for Claude Code, Claude Desktop, VS Code, Windsurf, Zed, Gemini CLI and the rest.
It listed its tools to an anonymous client, so reading the tool list needs no key. Individual tools can still ask for credentials in their own inputs.
Publisher: mcp.laisky.com. Endpoint host: mcp.laisky.com. ARD identifier: urn:air:mcp.laisky.com:mcp-server:laisky-mcp.
First indexed 18 September 2026. The tools are exactly what the server returned; nothing on this page
is inferred from its name. Ranking in Neuronto search is never for sale.