Skip to main content
Glama
Graphicle-AI

Graphicle MCP Server

Official

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
GRAPHICLE_NO_UPDATE_NOTIFIERNoSet to 1 to turn off the daily check for a new release, the only automatic request Graphicle makes.

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{}
resources
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
recall_projectB

Find original source evidence and current reviewed memory. Optional local model profiles select explicitly configured CPU embedding and reranking; leave the reranker unset for a project holding one conversation's sessions, where it did not raise held-out recall and took about 20 s per question. Continue incomplete sources before making decisions.

read_project_sourceB

Read an authorized original source by ID or repository path, with exact byte continuation.

repository_statusC

Inspect source coverage, source revisions and memory state.

repository_filesC

List authorized source paths and IDs.

project_memoryC

Read memory drafts, review state and freshness. Stored procedures are not execution permission.

search_project_memoryA

Search reviewed structured memory by its declared name or alias, scope and validity. Explicit lexical mode discovers text matches. Returns source citations and ambiguity diagnostics; makes no model call.

project_memory_historyA

Inspect immutable memory revisions, optionally as known at an RFC3339 timestamp.

prepare_memory_refreshC

Prepare exact current evidence and a stable document delta contract. No model call or memory write.

workflow_statusB

Inspect the caller's queued work and recorded model usage.

graph_episodesC

List recorded graph episodes and their labels for a repository.

graph_as_ofA

Query either valid_at/known_at/branch for independent world and recording time, or the repository's typed code graph at an episode selector (as_of: episode id, commit sha label, latest or working), reconstructed from recorded history without re-extraction. Narrow with node (exact id), query (id substring) or relations; includes parent episode, labels and cardinality-one contradictions recorded at that episode.

graph_linkA

Resolve imports between the independently indexed repositories of this workspace (Python, TypeScript/JavaScript, Go) by package identity and persist the cross-repository edges with both episodes; re-running after new episodes closes stale edges. Then recompute and persist the links from documents to the code symbols they name. Only repositories the caller may read take part.

graph_history_statsA

Storage and timing summary for a repository's graph history: row counts and content bytes per table (nodes, edges, edge_sources, coverage, cross_edges, unresolved_calls), episode count, recent coverage uploads with their format and granularity, the latest change-impact report's wall-clock time, and the retention defaults in force.

prepare_observationsC

Prepare current source excerpts, existing claim targets and the interpretation contract. Use your own model; no provider call.

session_timelineA

Browse captured coding-agent turns (Claude Code, Codex) of this project: the latest session by default, or the session of an anchor turn, with turns before and after it and the recent sessions. Returns titles, times and source ids; read_project_source returns a whole turn.

query_graphB

Search the knowledge graph using BFS or DFS. Returns relevant nodes and edges as text context.

get_nodeC

Get full details for a specific node by label or ID.

get_neighborsC

Get all direct neighbors of a node with edge details.

get_communityB

Get all nodes in a community by community ID.

god_nodesB

Return the most connected nodes - the core abstractions of the knowledge graph.

graph_statsB

Return summary statistics: node count, edge count, communities, confidence breakdown.

shortest_pathA

Find the shortest path between two concepts in the knowledge graph. Follows stored edge direction by default; set undirected=true to ignore it.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription
Graph ReportFull GRAPH_REPORT.md
Graph StatsNode/edge/community counts and confidence breakdown
God NodesTop 10 most-connected nodes
Surprising ConnectionsCross-community surprising connections
Confidence AuditEXTRACTED/INFERRED/AMBIGUOUS edge breakdown
Suggested QuestionsSuggested questions for this codebase

TDQS

B3.1/5.0

Scored across 22 tools

Disambiguation3/5

Several graph-query tools (query_graph, graph_as_of, get_node, get_neighbors, shortest_path, graph_stats, graph_history_stats, graph_episodes) all retrieve graph information, and memory tools (recall_project, search_project_memory, project_memory, prepare_observations, prepare_memory_refresh) have related search/preparation roles. Descriptions help distinguish temporal vs current, summary vs detail, but an agent can still hesitate between stats variants and memory preparation tools.

Naming Consistency4/5

All names use snake_case, which is consistent and readable. However, the pattern mixes verb_noun (get_node, query_graph, prepare_observations) with noun phrases (god_nodes, graph_stats, project_memory, repository_status), so it is not a uniform verb_noun convention.

Tool Count3/5

22 tools is at the upper end of the moderate range (16-25 feels heavy). The surface covers several subsystems (graph query, memory, source, sessions, workflow), which justifies some breadth, but consolidation of overlapping query/stats tools could reduce the count.

Completeness4/5

The server covers graph querying, temporal episodes, memory reading/search/history, source access, session browsing, and preparation contracts. Minor gaps include no explicit commit/apply tool for prepared memory or observations, and no general repository/project listing, but core workflows appear supported.

Maintenance

ActivityMaintained
ResponsivenessResponsive