LibrarySAFE
MCP server for operating on Markdown knowledge bases
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
library-mcp is an MCP server for interacting with Markdown knowledge bases. Basically, folders that may or may not have subfolders, that include files with .md extension and start with metadata like:
title: My blog post tags:
- python
- programming
url: /my-blog-post ---
Yesterday I was dreaming about...
The typical workflow in the current verison is to retrieve recent content for a given tag, and then discuss using that tag:
Get the next 50 posts with tag "executive", then tell me what I should do about this problem I am running into: ...
You can also do the same but by date ranges:
Summarize the blog posts I wrote in the past year.
You might reasonably ask "why not just upload your entire blog into the context window?" and there are two places where this library outperforms that approach:
- My blog corpus is much larger than most model's context windows today.
Further, even if the context windows became exhaustively large, I wrote a lot of mediocre stuff in the past, so maybe omitting it's a feature.
- I have a number of distinct Markdown knowledge bases, and this lets me
operate across them in tandem.
Finally, this is a hobby project, intended for running locally on your laptop. No humans have been harmed using this software, but it does work pretty well!
Tools
This MCP server exposes these tools.
Content Search Tools
Tools for retrieving content into your context window:
get_by_tag- Retrieves content by tagget_by_text- Searches content for specific textget_by_slug_or_url- Finds posts by slug or URLget_by_date_range- Gets posts published within a date range
Tag Management Tools
Tools for navigating your knowledge base:
search_tags- Searches for tags matching a querylist_all_tags- Lists all tags sorted by post count and recency
Maintenance Tools
Tools for dealing with r
6f744145a749OBSERVED · 2026-10-07Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add library-mcp -- uvx library-mcp
{
"mcpServers": {
"library-mcp": {
"command": "uvx",
"args": [
"library-mcp"
]
}
}
}Exposed tools (7)
5 read · 2 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
get_by_date_range | read | Get posts published within a date range. |
get_by_slug_or_url | write | Get a post by its slug or URL. |
get_by_tag | read | Get blog content by its tag. |
get_by_text | read | Get blog content by text in content. |
list_all_tags | write | List all tags sorted by number of posts and most recent post. |
rebuild | read | Rebuild text index. Useful for when contents have changed on disk |
search_tags | read | Search for tags matching the provided query. |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (1)
Gates applied: no_behavioural_pass, no_license.
6f744145a749full audit observations/trust-audit/mcp-server/lethain__library.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-07 | 6f744145a749 | SAFE | B | 89 | first audit |
Questions
What is the Library MCP server?
MCP server for operating on Markdown knowledge bases
What tools does Library expose?
7 in total: 5 read-only, 2 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is Library safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does Library need?
No credential environment variables were found in its source, so it appears to need none.
How does Library run?
It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as library-mcp.
How current is this page?
The grade is for one exact copy of the source (6f744145a749), read on 2026-10-07. The repository is watched and re-audited when it changes.