codemem

name	codemem
description	Quick reference for all 32 codemem MCP tools. Use when working with the codemem knowledge graph — finding code, traversing relationships, storing memories, or running analysis.
user-invocable	true
argument-hint	[query or topic]

Codemem Tool Guide

Quick reference for codemem's 32 MCP tools. If arguments were provided, use them as context for the most relevant tool below.

$ARGUMENTS

Finding Code & Symbols

Scenario	Tool	Key Params
Find function by name	search_code	query, mode: "text"
Find code by concept	search_code	query, mode: "semantic"
Find code by name + concept	search_code	query, mode: "hybrid"
Get full symbol details	get_symbol_info	qualified_name
Browse file/package tree	summary_tree	start_id: "pkg:src/", max_depth

search_code also accepts: k (result count, default 10), kind (filter by symbol kind).

Graph Traversal

Scenario	Tool	Key Params
What calls this function?	get_symbol_graph	qualified_name, direction: "incoming", depth: 1
What does this function call?	get_symbol_graph	qualified_name, direction: "outgoing", depth: 1
Full blast radius of a change	get_symbol_graph	qualified_name, direction: "incoming", depth: 2
Walk relationships from any node	graph_traverse	start_id, max_depth, include_relationships, exclude_kinds
Find most critical symbols	find_important_nodes	top_k: 20 (PageRank, damping: 0.85)
Find related symbol clusters	find_related_groups	resolution: 1.0 (Louvain communities)

graph_traverse also accepts: algorithm ("bfs"/"dfs"), exclude_kinds, include_relationships.

get_symbol_graph directions: "incoming", "outgoing", "both" (default).

Memories (Stored Knowledge)

Scenario	Tool	Key Params
Ask a question about the codebase	recall	query
Ask with graph context expansion	recall	query, expand: true, expansion_depth: 2
Ask with architectural impact data	recall	query, include_impact: true
Get memories linked to a symbol/file	get_node_memories	node_id: "sym:Name" or "file:path"
Check if symbols have documentation	node_coverage	node_ids: ["sym:X", "sym:Y"]
Follow decision evolution over time	get_decision_chain	topic and/or file_path
Store a finding	store_memory	content, memory_type, importance, tags, links
Update a finding (new version)	refine_memory	id, content (creates EVOLVED_INTO edge)
Update in-place (destructive)	refine_memory	id, content, destructive: true
Split a large memory	split_memory	id, parts (creates PART_OF edges)
Merge related memories	merge_memories	ids, content (creates SUMMARIZES edges)
Delete a memory	delete_memory	id
Link two memories	associate_memories	source_id, target_id, relationship

recall also accepts: k, memory_type, namespace, exclude_tags, min_importance, min_confidence.

store_memory types: decision, pattern, preference, style, habit, insight, context (default).

Analysis & Health

Scenario	Tool	Key Params
Check graph size & health	codemem_status	include: ["stats", "health", "metrics"]
Detect recurring patterns	detect_patterns	min_frequency: 3
Full pipeline (index+enrich+rank)	CLI: codemem analyze	--skip-enrich, --skip-embed, --force
Deduplicate similar memories	consolidate	mode: "cluster", similarity_threshold: 0.85
Clean up low-value memories	consolidate	mode: "forget", importance_threshold: 0.3
Find creative cross-connections	consolidate	mode: "creative"
Decay old memories	consolidate	mode: "decay", threshold_days: 30
Summarize memory clusters	consolidate	mode: "summarize", cluster_size: 5
Auto-select consolidation	consolidate	mode: "auto"

Namespace & Session Management

Scenario	Tool	Key Params
List all namespaces	list_namespaces	(no params)
Namespace stats	namespace_stats	namespace
Delete a namespace	delete_namespace	namespace
Session progress snapshot	session_checkpoint	session_id
Get session context	session_context	namespace, k
Cross-repo memories	get_cross_repo	namespace

Quick Decision Tree

• "What calls X / what does X depend on?" -> get_symbol_graph
• "How does feature Y work?" -> recall + search_code mode=semantic
• "Is it safe to change X?" -> get_symbol_graph direction=incoming depth=2
• "Do we already have something like Z?" -> search_code mode=semantic
• "Why was X designed this way?" -> recall or get_node_memories
• "Which files matter most?" -> find_important_nodes
• "What's related to what?" -> find_related_groups
• "Has this been documented?" -> node_coverage
• "What changed recently?" -> recall query="pending analysis"

Relationship Types (for associate_memories)

RELATES_TO, LEADS_TO, PART_OF, REINFORCES, CONTRADICTS, EVOLVED_INTO, DERIVED_FROM, INVALIDATED_BY, DEPENDS_ON, IMPORTS, EXTENDS, CALLS, CONTAINS, SUPERSEDES, BLOCKS, IMPLEMENTS, INHERITS, SIMILAR_TO, PRECEDED_BY, EXEMPLIFIES, EXPLAINS, SHARES_THEME, SUMMARIZES, CO_CHANGED