Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
ANTHROPIC_API_KEYNoAPI key for Claude (Anthropic) for synthesis. If not set, uses local Ollama model.

Capabilities

Features and capabilities supported by this server

CapabilityDetails
tools
{
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
search_vaultA

Search your personal knowledge vault using hybrid retrieval (semantic + keyword + knowledge graph). Returns cited passages from your notes.

get_entity_neighborsA

Given a concept or entity name, find everything connected to it in the knowledge graph. Returns related entities and the typed relationships between them (CITES, INFLUENCES, EXTENDS, CONTRASTS_WITH, etc) with the evidence spans that justify each edge.

get_archaeology_reportC

Surface forgotten connections in your vault. Finds notes and concepts that are semantically related but never linked, and clusters of knowledge you may have forgotten about.

ingest_urlA

Add a web URL to your vault. Loom will fetch, parse, and index it. Supports general web pages, arXiv papers, Wikipedia articles, HN threads, and more.

get_vault_statsA

Get statistics about your knowledge vault: total documents, chunks indexed, entity count, and top concepts by frequency.

get_vault_healthA

Diagnose vault hygiene: orphan notes (no wikilinks in or out), broken wikilinks (point to missing files), stale notes (old and weakly connected), and tag issues (likely typos, inconsistent casing). Works in vector-only mode; uses the graph for connectivity when Neo4j is available.

run_loopA

Trigger a Loom background loop on demand. 'vault_health' writes a health report to the vault; 'stale_check' writes a stale-notes review; 'index_freshness' re-indexes files changed since the last index pass; 'all' runs every loop that is due. Reports are written into the vault's 'Loom Reports' folder.

create_noteA

Create a new markdown note in the vault. Refuses to overwrite existing files. Adds YAML frontmatter (source: mcp_create, plus any tags). Smart-write pipeline: suggests existing tags for near-duplicates, discovers wikilink candidates among existing filenames, and — when 'folder' is omitted — places the note in the folder of the most semantically similar existing note.

edit_sectionA

Edit one section of an existing note, located by heading (any level, fuzzy filename match). mode = replace | prepend | append. All content outside the targeted section is preserved. Returns the edited section for confirmation.

read_sectionA

Read just one section of a note (from a heading to the next heading of equal or higher level). Fuzzy filename match. Lower token cost than reading the whole document.

append_to_noteA

Append content to the end of an existing vault note (fuzzy filename match). Simpler than edit_section — no heading needed.

add_factA

Write a correction into the knowledge graph. With only 'subject', declares/upserts a single entity. With 'subject', 'relation', and 'object', asserts a typed relationship between two entities (creating either endpoint if missing). Valid relations: CITES, CONTRASTS_WITH, EXTENDS, INFLUENCES, UNCERTAIN_SAME_AS, UNTYPED_RELATION. Manual writes are tagged source='manual_mcp'. Use for surgical fixes, not bulk import.

delete_edgeA

Remove a single relationship from the knowledge graph by its edge_id. Get edge_ids from get_entity_neighbors. Deletes only the edge — both entities remain.

forget_entityA

Remove an entity and ALL of its relationships from the knowledge graph (and any aliases left orphaned). Matches by exact canonical key — surgical, not fuzzy. Destructive and not reversible without re-indexing.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

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/KlossKarl/loom'

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