Install

npm install -g cogni-code
cogni-code install

Detects every installed harness (Claude Code, Codex CLI, OpenCode) and wires them all. Memory is shared across harnesses — install into several and they all see the same graph.

For a custom storage location: cogni-code install --graph-root /path/to/memory

npm install -g cogni-code
cogni-code install --claude

Symlinks the package into ~/.claude/plugins/. Restart Claude Code or run /mcp to reconnect — the graph_memory tool and all slash commands come live.

Does not touch settings.json.

npm install -g cogni-code
cogni-code install --codex

Merges hooks into ~/.codex/hooks.json using CLI commands (not absolute paths, so updates propagate automatically). Run /hooks in Codex to review and trust the cogni-code hooks.

npm install -g cogni-code
cogni-code install --opencode

Installs the native plugin extension and slash commands into your OpenCode config directory.

Locked-down containers may block npm install -g. Use a local install:

npm install cogni-code
npx cogni-code install

The container likely has no Docker. Set up the pipeline with the api worker (see API mode below) — it calls the Anthropic API directly via fetch and routes through your existing credential proxy (e.g. OneCLI Agent Vault) for subscription access.

Permission denied (EACCES)?

Use a local install instead of global:

npm install cogni-code && npx cogni-code install

All commands work the same — just prefix with npx.

Updates are automatic — npm update -g cogni-code updates all hooks instantly (they call the cogni-code CLI, not absolute file paths).

Background pipeline

Docker mode (recommended for desktops)

cogni-code install --docker

Sets up the Docker daemon that runs the scribe/librarian/dreamer pipeline automatically. Requires Docker Desktop or Podman. Auto-detects the worker provider (codex, claude, opencode, or api) or specify with --worker codex.

API mode (no Docker required)

If your environment has no Docker or Podman — a NanoClaw container, a sandbox, a CI runner — the api worker runs the full pipeline via direct HTTP calls. No CLI agent binary, no subprocess:

cogni-code install --docker --worker api

The api worker respects the standard Anthropic env-var surface, so it routes through whatever credential infrastructure your agent already uses:

Env var	Auth method	Use case
`ANTHROPIC_API_KEY`	`x-api-key` header	Direct API key (bills per-token)
`ANTHROPIC_AUTH_TOKEN` + `ANTHROPIC_BASE_URL`	`Authorization: Bearer`	Credential proxy (e.g. OneCLI Agent Vault in NanoClaw) — routes to your subscription, no API billing

It auto-activates when credentials are present and no CLI harness is detected. You can also start the daemon directly without Docker:

node "$(npm root -g)/cogni-code/dist/graph-memory/pipeline/daemon.js"

From source (git clone)

git clone https://github.com/ConnorCallahan01/cogni-code.git
cd cogni-code/graph-memory-plugin
npm install && npm run build

./bin/install.sh           # Claude Code
./bin/install-codex.sh     # Codex CLI
./bin/install-opencode.sh  # OpenCode

First run

Then start a session and run:

/memory-onboard

The onboard wizard walks you through runtime mode and seeds your first memory nodes. If you installed via npm, the graph is already initialized — onboarding is optional.

Installing as an agent? See the agent install guide — a flat, executable runbook designed for AI agents given a URL and told to install.