Use for operator-grade Gingko memory operations the daily CLI does not expose. Triggers on "fork branch memory", "merge branch memory", "list memory branches", "prune branch", "set charter", "project charter", "force commit session", "flush memory now", "run maintenance", "decay memories", "consolidate memories", "refresh project summary", "get cluster". Only invoke when the user explicitly asks.
Use to search past Gingko memories before starting work, when the user says "what did we decide", "have we seen this", "look this up", "did we already", "again", "still", "back to", or when about to make an architectural call in a Gingko-enabled project (SessionStart prints "[gingko] primed session context"). Use any time prior sessions may already contain the answer.
Use to save findings to Gingko memory after a bug fix, non-obvious decision, meaningful refactor, library choice, schema/migration decision, or anything surprising. Triggers on "remember that", "record this", "save to memory", "save what we learned", "note this for later", or after completing a task whose reasoning is worth keeping in a Gingko-enabled project.
Use when Gingko memory misbehaves. Triggers on "gingko isn't working", "memory broken", "service unreachable", "no active session", "session-id is empty", "recall returns nothing", "gingko: command not found", "append-step failing", "wrong project id", "stale primer", or any user complaint about Gingko hooks/CLI/MCP failing.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Gingko is a application that exposes a project-scoped memory graph over MCP. Any agent that speaks MCP can open a project, record observations, recall past memories, and navigate the knowledge graph that builds up across sessions.
Under the hood it uses Mnemosyne as a memory engine
For development, if you have mise, just run:
mix setup # fetch deps, install assets
mix phx.server
Then point an MCP client at http://localhost:8008/mcp. The first time the
server boots it creates ~/.gingko/ with a default config and guides you
through setup at http://localhost:8008/setup if anything is missing.
Long-form guides live in guides/:
config.tomlWrite flow:
open_project_memory → start_session → append_step → (auto-commit on
session end; close_async or commit_session for explicit flushes)Read flow:
recall, get_node, get_session_state, list_projects, latest_memoriesSummary layer (opt-out via [summaries].enabled):
get_session_primer, get_cluster, set_charter, refresh_principal_memoryGraph maintenance:
run_maintenance with decay, consolidate, or validateSee MCP Tools for the full reference.
All state lives under $GINGKO_HOME (default ~/.gingko):
~/.gingko/
├── config.toml # runtime configuration
├── memory/ # Mnemosyne DETS graph files, one subdir per project
└── metadata.sqlite3 # projects, sessions, summaries
API keys are never written to config.toml — only the env-var names to read.
Export ANTHROPIC_API_KEY, OPENAI_API_KEY, or whichever provider keys your
configured providers need.
Asset toolchain requirement: Yarn 1.x, or Node with Corepack enabled.
mix setup auto-detects yarn, then corepack yarn, and exits with a clear
message if neither is available.
mix setup
mix phx.server # http://localhost:8008
The web UI has three main views:
/ — project grid with live stats./setup — edit config.toml from the browser./projects — real-time graph monitor backed by Cytoscape.mix test # full suite
mix precommit # warnings-as-errors + deps.unlock --unused + format + test
Integration coverage for the MCP tools lives in
test/gingko/mcp/write_flow_test.exs and test/gingko/mcp/read_flow_test.exs.
npx claudepluginhub edlontech/edlon-agents-marketplace --plugin gingkoMemory compression system for Claude Code - persist context across sessions
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Streamline people operations — recruiting, onboarding, performance reviews, compensation analysis, and policy guidance. Maintain compliance and keep your team running smoothly.
Write feature specs, plan roadmaps, and synthesize user research faster. Keep stakeholders updated and stay ahead of the competitive landscape.