Skip to content

CLI Reference

Complete reference for the nmem command-line interface. 82 commands available.

Tip

Run nmem --help or nmem <command> --help for the latest usage info.

Table of Contents

Memory Operations

nmem remember

Store a new memory (type auto-detected if not specified).

nmem remember [OPTIONS]
Option Type Required Default Description
content text No `` (positional argument)
--tag / -t text No — Tags for the memory
--type / -T text No — Memory type: fact, decision, preference, todo, insight, context, instruction, error, workflow, reference (auto-detect...
--priority / -p integer No — Priority 0-10 (0=lowest, 5=normal, 10=critical)
--expires / -e integer No — Days until this memory expires
--project / -P text No — Associate with a project (by name)
--shared / -S boolean No False Use shared/remote storage for this command
--force / -f boolean No False Store even if sensitive content is detected or encoding damage is suspected
--redact / -r boolean No False Auto-redact sensitive content before storing
--timestamp / --at text No — ISO datetime of original event (e.g. '2026-03-02T08:00:00'). Defaults to now.
--ephemeral boolean No False Session-scoped memory: auto-expires after 24h, never synced
--stdin boolean No False Read content from stdin even when an argument is also present
--file text No — Read content from a UTF-8 file (bypasses shell pipe encoding entirely)
--json / -j boolean No False Output as JSON

A pipe is decoded as UTF-8 when the bytes are valid UTF-8, otherwise with the Windows OEM code page. PowerShell 5.1 pipes ASCII and replaces CJK with "?". CP1252 cannot represent CJK either. Use --file for those.

nmem recall

Query memories with intelligent routing (query type auto-detected).

nmem recall [OPTIONS]
Option Type Required Default Description
query text Yes — (positional argument)
--depth / -d integer No — Search depth (0=instant, 1=context, 2=habit, 3=deep)
--limit / -l integer No — Cap matched memories (approximate; maps to max_tokens ~200/memory)
--max-tokens / -m integer No 500 Max tokens in response
--min-confidence / -c float No 0.0 Minimum confidence threshold (0.0-1.0)
--shared / -S boolean No False Use shared/remote storage for this command
--show-age / -a boolean No True Show memory ages in results
--show-routing / -R boolean No False Show query routing info
--json / -j boolean No False Output as JSON

nmem context

Get recent context (for injecting into AI conversations).

nmem context [OPTIONS]
Option Type Required Default Description
--limit / -l integer No 10 Number of recent memories
--fresh-only boolean No False Only include memories < 30 days old
--json / -j boolean No False Output as JSON

nmem todo

Quick shortcut to add a TODO memory.

nmem todo [OPTIONS]
Option Type Required Default Description
task text Yes — (positional argument)
--priority / -p integer No 5 Priority 0-10 (default: 5=normal, 7=high, 10=critical)
--project / -P text No — Associate with a project
--expires / -e integer No — Days until expiry (default: 30)
--tag / -t text No — Tags for the task
--json / -j boolean No False Output as JSON

nmem q

Quick recall - shortcut for 'nmem recall'.

nmem q [OPTIONS]
Option Type Required Default Description
query text Yes — (positional argument)
-d integer No — —

nmem a

Quick add - shortcut for 'nmem remember' with auto-detect.

nmem a [OPTIONS]
Option Type Required Default Description
content text Yes — (positional argument)
-p integer No — —

nmem last

Show last N memories - quick view of recent activity.

nmem last [OPTIONS]
Option Type Required Default Description
-n integer No 5 Number of memories to show

nmem today

Show today's memories.

nmem today [OPTIONS]

Brain Management

nmem brain list

List available brains.

nmem brain list [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

nmem brain use

Switch to a different brain.

nmem brain use [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)

This updates config.toml so CLI and MCP servers (without NMEM_BRAIN env var) will use the new brain. MCP servers started with NMEM_BRAIN are pinned to that brain and will NOT be affected by this command.

nmem brain create

Create a new brain.

nmem brain create [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--use / -u boolean No True Switch to the new brain after creating

nmem brain export

Export brain to JSON or markdown file.

nmem brain export [OPTIONS]
Option Type Required Default Description
--output / -o text No — Output file path
--name / -n text No — Brain name (default: current)
--exclude-sensitive / -s boolean No False Exclude memories with sensitive content
--format / -f text No json Export format: json or markdown

nmem brain import

Import brain from JSON file.

nmem brain import [OPTIONS]
Option Type Required Default Description
file text Yes — (positional argument)
--name / -n text No — Name for imported brain
--use / -u boolean No True Switch to imported brain
--scan boolean No True Scan for sensitive content before importing

nmem brain delete

Delete a brain.

nmem brain delete [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--force / -f boolean No False Skip confirmation

nmem brain health

Check brain health (freshness, sensitive content).

nmem brain health [OPTIONS]
Option Type Required Default Description
--name / -n text No — Brain name (default: current)
--json / -j boolean No False Output as JSON

nmem brain transplant

Transplant memories from another brain into the current brain.

nmem brain transplant [OPTIONS]
Option Type Required Default Description
source text Yes — (positional argument)
--tag / -t text No — Filter by tags
--type text No — Filter by memory types
--strategy / -s text No prefer_local Conflict resolution strategy
--json / -j boolean No False Output as JSON

Information & Diagnostics

nmem stats

Show brain statistics including freshness and memory type analysis.

nmem stats [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

nmem status

Show current brain status, recent activity, and actionable suggestions.

nmem status [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

Unlike 'stats' which shows raw counts, 'status' gives a quick overview of what's happening and what you should do next.

nmem health

Show brain health diagnostics with purity score and recommendations.

nmem health [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

Analyzes connectivity, diversity, freshness, consolidation, orphan rate, and activation efficiency to produce a composite health grade (A-F) and actionable recommendations.

nmem check

Check content for sensitive information without storing.

nmem check [OPTIONS]
Option Type Required Default Description
content text Yes — (positional argument)
--json / -j boolean No False Output as JSON

nmem doctor

Run system health diagnostics.

nmem doctor [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON
--fix boolean No False Auto-fix available issues
--dev boolean No False Include source checkout and contributor tooling checks

Checks Python version, config, brain, dependencies, embedding provider, schema version, MCP config, hooks, dedup, surface, and CLI tools.

Use --fix to automatically remediate fixable issues. Use --dev to check contributor setup in a source checkout.

nmem dashboard

Show a rich dashboard with brain stats and recent activity.

nmem dashboard [OPTIONS]

Displays: - Brain statistics (neurons, synapses, fibers) - Memory types distribution - Freshness analysis - Recent memories

nmem ui

Interactive memory browser with rich formatting.

nmem ui [OPTIONS]
Option Type Required Default Description
--type / -t text No — Filter by memory type
--search / -s text No — Search in memory content
--limit / -n integer No 20 Number of memories to show

Browse memories with color-coded types, priorities, and freshness.

nmem graph

Visualize neural connections as a tree graph.

nmem graph [OPTIONS]
Option Type Required Default Description
query text No — (positional argument)
--depth / -d integer No 2 Traversal depth (1-3)
--export / -e text No — Export format: svg
--output / -o text No — Output file path (used with --export)

Shows memories and their relationships (caused_by, leads_to, etc.)

Training & Import/Export

nmem train

Train a brain from documentation files (markdown).

nmem train [OPTIONS]
Option Type Required Default Description
path text No . (positional argument)
--domain / -d text No `` Domain tag (e.g., react, kubernetes)
--brain / -b text No `` Target brain name (default: current)
--ext / -e text No — File extensions (default: .md)
--no-consolidate boolean No False Skip ENRICH consolidation
--json / -j boolean No False Output as JSON

nmem index

Index a codebase into neural memory for code-aware recall.

nmem index [OPTIONS]
Option Type Required Default Description
path text No . (positional argument)
--ext / -e text No — File extensions to index (e.g. .py)
--status / -s boolean No False Show indexing status instead of scanning
--json / -j boolean No False Output as JSON

nmem import

Import brain from JSON file.

nmem import [OPTIONS]
Option Type Required Default Description
input_file text Yes — (positional argument)
--brain / -b text No — Target brain name (default: from file)
--merge / -m boolean No False Merge with existing brain
--strategy text No prefer_local Conflict resolution: prefer_local, prefer_remote, prefer_recent, prefer_stronger

nmem export

Export brain to JSON file for backup or sharing.

nmem export [OPTIONS]
Option Type Required Default Description
output text Yes — (positional argument)
--brain / -b text No — Brain to export (default: current)

Configuration & Setup

nmem init

Set up NeuralMemory in one command.

nmem init [OPTIONS]
Option Type Required Default Description
--force / -f boolean No False Overwrite existing config
--skip-mcp boolean No False Skip MCP auto-configuration
--skip-skills boolean No False Skip skills installation
--wizard / -w boolean No False Interactive setup wizard
--defaults boolean No False Non-interactive with all defaults
--full boolean No False Extended setup: embeddings, dedup, maintenance script
--embeddings boolean No False Set up embedding provider
--skip-embeddings boolean No False Skip embedding provider setup
--dedup boolean No False Enable dedup

Creates config, default brain, auto-configures MCP for Claude Code and Cursor, and installs agent skills.

Use --full for extended setup: auto-detect embeddings, enable dedup, and generate a maintenance script.

nmem setup

Set up optional components.

nmem setup [OPTIONS]
Option Type Required Default Description
component text No `` (positional argument)
--ide text No `` Target IDE for rules: cursor, windsurf, cline, gemini, agents
--all boolean No False Generate rules for all supported IDEs
--force / -f boolean No False Overwrite existing files

nmem mcp-config

Generate MCP server configuration for Claude Code/Cursor.

nmem mcp-config [OPTIONS]
Option Type Required Default Description
--with-prompt / -p boolean No False Include system prompt in config
--compact / -c boolean No False Use compact prompt (if --with-prompt)

Outputs JSON configuration that can be added to your MCP settings.

nmem prompt

Show system prompt for AI tools.

nmem prompt [OPTIONS]
Option Type Required Default Description
--compact / -c boolean No False Show compact version
--copy boolean No False Copy to clipboard (requires pyperclip)

This prompt instructs AI assistants (Claude, GPT, etc.) on when and how to use NeuralMemory for persistent memory across sessions.

nmem hooks

Install or manage git hooks for automatic memory capture.

nmem hooks [OPTIONS]
Option Type Required Default Description
action text No install (positional argument)
--path / -p text No — Path to git repo (default: current dir)

nmem config preset

Apply a configuration preset or list available presets.

nmem config preset [OPTIONS]
Option Type Required Default Description
name text No `` (positional argument)
--list / -l boolean No False List available presets
--dry-run / -n boolean No False Show changes without applying

nmem config tier

Get or set the MCP tool tier to control token usage.

nmem config tier [OPTIONS]
Option Type Required Default Description
name text No `` (positional argument)
--show / -s boolean No False Show current tier

Tiers control which tools are exposed in tools/list: minimal — 4 core tools (~84% token savings) standard — 8 tools (~69% savings) full — all 63 tools (default)

Hidden tools remain callable — only schema exposure changes.

nmem install-skills

Install NeuralMemory skills to ~/.claude/skills/.

nmem install-skills [OPTIONS]
Option Type Required Default Description
--force / -f boolean No False Overwrite existing skills with latest version
--list / -l boolean No False List available skills without installing

Skills add AI agent workflows for memory management: memory-intake Structured memory creation from messy notes memory-audit 6-dimension quality review with graded findings memory-evolution Evidence-based optimization from usage patterns

Server & MCP

nmem serve

Run the NeuralMemory API server.

nmem serve [OPTIONS]
Option Type Required Default Description
--host / -h text No 127.0.0.1 Host to bind to
--port / -p integer No 8000 Port to bind to
--reload / -r boolean No False Enable auto-reload for development

nmem mcp

Run the MCP (Model Context Protocol) server.

nmem mcp [OPTIONS]

This starts an MCP server over stdio that exposes NeuralMemory tools to Claude Code, Claude Desktop, and other MCP-compatible clients.

Available tools: nmem_remember - Store a memory nmem_recall - Query memories nmem_context - Get recent context nmem_todo - Add a TODO memory nmem_stats - Get brain statistics

Configuration for Claude Code: claude mcp add --scope user neural-memory -- nmem-mcp

Maintenance

nmem decay

Apply memory decay to simulate forgetting.

nmem decay [OPTIONS]
Option Type Required Default Description
--brain / -b text No — Brain to apply decay to
--dry-run / -n boolean No False Preview changes without applying
--prune / -p float No 0.01 Prune below this activation level

Memories that haven't been accessed recently will have their activation levels reduced following the Ebbinghaus forgetting curve.

nmem consolidate

Consolidate brain memories by pruning, merging, or summarizing.

nmem consolidate [OPTIONS]
Option Type Required Default Description
--brain / -b text No — Brain to consolidate
--strategy / -s text No all Consolidation strategy. Valid values: prune, merge, summarize, mature, infer, enrich, dream, learn_habits, dedup, sem...
--dry-run / -n boolean No False Preview changes without applying
--prune-threshold float No 0.05 Weight threshold for pruning synapses
--merge-overlap float No 0.5 Jaccard overlap threshold for merging fibers
--min-inactive-days float No 7.0 Minimum inactive days before pruning

Runs one or more consolidation passes on the current (or specified) brain. Use --strategy to pick a specific pass, or omit it to run everything.

Strategies: prune - Remove weak synapses and orphan neurons merge - Combine overlapping fibers summarize - Create concept neurons for topic clusters mature - Advance episodic memories to semantic stage infer - Add inferred synapses from co-activation patterns enrich - Enrich neurons with extracted metadata dream - Generate synthetic bridging memories learn_habits - Extract recurring workflow patterns dedup - Merge near-duplicate memories semantic_link - Add cross-domain semantic connections compress - Compress old low-activation fibers all - Run all strategies in dependency order (default)

nmem cleanup

Clean up expired or old memories.

nmem cleanup [OPTIONS]
Option Type Required Default Description
--expired / -e boolean No True Only clean up expired memories
--type / -T text No — Only clean up specific memory type
--dry-run / -n boolean No False Show what would be deleted without deleting
--force / -f boolean No False Skip confirmation
--json / -j boolean No False Output as JSON

nmem flush

Emergency flush: capture memories before context is lost.

nmem flush [OPTIONS]
Option Type Required Default Description
--transcript / -t text No — Path to JSONL transcript file
text text No — (positional argument)
--json / -j boolean No False Output as JSON

Aggressively detects and saves all memorable content with lowered confidence threshold. Designed for pre-compaction or session-end scenarios.

Project Management

nmem project create

Create a new project for organizing memories.

nmem project create [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--description / -d text No — Project description
--duration / -D integer No — Duration in days (creates end date)
--tag / -t text No — Project tags
--priority / -p float No 1.0 Project priority (default: 1.0)
--json / -j boolean No False Output as JSON

Projects group related memories and enable time-bounded retrieval.

nmem project list

List all projects.

nmem project list [OPTIONS]
Option Type Required Default Description
--active / -a boolean No False Show only active projects
--json / -j boolean No False Output as JSON

nmem project show

Show project details and its memories.

nmem project show [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--json / -j boolean No False Output as JSON

nmem project delete

Delete a project (memories are preserved but unlinked).

nmem project delete [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--force / -f boolean No False Skip confirmation

nmem project extend

Extend a project's deadline.

nmem project extend [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
days integer Yes — (positional argument)
--json / -j boolean No False Output as JSON

Advanced Features

nmem shared enable

Enable shared mode to connect to a remote NeuralMemory server.

nmem shared enable [OPTIONS]
Option Type Required Default Description
server_url text Yes — (positional argument)
--api-key / -k text No — API key for authentication
--timeout / -t float No 30.0 Request timeout in seconds

When shared mode is enabled, all memory operations (remember, recall, etc.) will use the remote server instead of local storage.

nmem shared disable

Disable shared mode and use local storage.

nmem shared disable [OPTIONS]

nmem shared status

Show shared mode status and configuration.

nmem shared status [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

nmem shared test

Test connection to the shared server.

nmem shared test [OPTIONS]

nmem shared sync

Manually sync local brain with remote server.

nmem shared sync [OPTIONS]
Option Type Required Default Description
--direction / -d text No both Sync direction: push, pull, or both
--json / -j boolean No False Output as JSON

Directions: push - Upload local brain to server pull - Download brain from server to local both - Full bidirectional sync (default)

nmem habits list

List learned workflow habits.

nmem habits list [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

Shows all habits discovered through action sequence mining, including step sequences, frequencies, and confidence scores.

nmem habits show

Show details of a specific learned habit.

nmem habits show [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--json / -j boolean No False Output as JSON

nmem habits clear

Clear all learned habits.

nmem habits clear [OPTIONS]
Option Type Required Default Description
--force / -f boolean No False Skip confirmation

Removes all WORKFLOW fibers with _habit_pattern metadata. This does not affect action event history.

nmem habits status

Show progress toward habit detection.

nmem habits status [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

Displays emerging action patterns that haven't yet reached the frequency threshold to become learned habits. Helps you understand how close your workflows are to being recognized.

nmem version create

Create a version snapshot of the current brain state.

nmem version create [OPTIONS]
Option Type Required Default Description
name text Yes — (positional argument)
--description / -d text No `` Description
--json / -j boolean No False Output as JSON

nmem version list

List brain versions.

nmem version list [OPTIONS]
Option Type Required Default Description
--limit / -l integer No 20 Max versions
--json / -j boolean No False Output as JSON

nmem version rollback

Rollback brain to a previous version.

nmem version rollback [OPTIONS]
Option Type Required Default Description
version_id text Yes — (positional argument)
--json / -j boolean No False Output as JSON

nmem version diff

Compare two brain versions.

nmem version diff [OPTIONS]
Option Type Required Default Description
from_version text Yes — (positional argument)
to_version text Yes — (positional argument)
--json / -j boolean No False Output as JSON

nmem telegram status

Show Telegram integration status.

nmem telegram status [OPTIONS]

nmem telegram test

Send a test message to verify configuration.

nmem telegram test [OPTIONS]

nmem telegram backup

Send brain database file as backup to Telegram.

nmem telegram backup [OPTIONS]
Option Type Required Default Description
--brain / -b text No — Brain name (default: active brain)

nmem list

List memories with filtering by type, priority, project, and status.

nmem list [OPTIONS]
Option Type Required Default Description
--type / -T text No — Filter by memory type (fact, decision, todo, etc.)
--min-priority / -p integer No — Minimum priority (0-10)
--project / -P text No — Filter by project name
--expired / -e boolean No False Show only expired memories
--include-expired boolean No False Include expired memories in results
--limit / -l integer No 20 Maximum number of results
--full boolean No False Return full content instead of a 100-char summary
--json / -j boolean No False Output as JSON

Content is summarised to 100 characters by default; pass --full (or use nmem show <id>) when the complete text matters.

Memory types: fact, decision, preference, todo, insight, context, instruction, error, workflow, reference

nmem migrate

Migrate brain data between storage backends.

nmem migrate [OPTIONS]
Option Type Required Default Description
target text Yes — (positional argument)
--brain / -b text No — Specific brain to migrate (default: current)
--pg-host text No localhost PostgreSQL host
--pg-port integer No 5432 PostgreSQL port
--pg-database text No neuralmemory PostgreSQL database name
--pg-user text No postgres PostgreSQL user
--pg-password text No `` PostgreSQL password (or use NEURAL_MEMORY_POSTGRES_PASSWORD env)

nmem update

Update neural-memory to the latest version.

nmem update [OPTIONS]
Option Type Required Default Description
--force / -f boolean No False Force update even if already latest
--check / -c boolean No False Only check for updates, don't install

Automatically detects install method (pip or git source) and updates accordingly.

Other

nmem edit

Edit a memory's type, content, priority, or tier.

nmem edit [OPTIONS]
Option Type Required Default Description
memory_id text Yes — (positional argument)
--type / -T text No — New memory type
--content / -c text No — New content for the anchor neuron
--priority / -p integer No — New priority 0-10
--tier text No — New tier: hot, warm, or cold
--shared / -S boolean No False Use shared/remote storage
--json / -j boolean No False Output as JSON

nmem forget

Soft- or hard-delete a memory by fiber ID (or neuron ID with --hard).

nmem forget [OPTIONS]
Option Type Required Default Description
memory_id text Yes — (positional argument)
--hard boolean No False Permanent deletion (cascade cleanup). Default is soft delete (expire now).
--reason / -r text No `` Reason for forgetting (logged for audit)
--force / -f boolean No False Confirm permanent deletion when using --hard
--shared / -S boolean No False Use shared/remote storage
--json / -j boolean No False Output as JSON

Soft delete sets expires_at=now; consolidation removes the fiber later. Hard delete drops the fiber + typed_memory immediately (CASCADE handles fiber_neurons) and requires --force, mirroring the confirmation gate on nmem_cleanup and on the MCP tools.

nmem lifecycle

Manage memory lifecycle states.

nmem lifecycle [OPTIONS]
Option Type Required Default Description
action text No status (positional argument)
neuron_id text No — (positional argument)
--brain / -b text No — Brain name

View lifecycle distribution or freeze/thaw/recover individual memories.

nmem pin

Pin a fiber as permanent KB (skips decay/prune; promotes tier to HOT).

nmem pin [OPTIONS]
Option Type Required Default Description
fiber_id text Yes — (positional argument)
--shared / -S boolean No False Use shared/remote storage
--limit / -l integer No 50 When fiber_id='list', cap number of results
--json / -j boolean No False Output as JSON

Use 'list' as the argument to list currently-pinned fibers.

nmem shared activate

Activate a NeuralMemory Pro license key.

nmem shared activate [OPTIONS]
Option Type Required Default Description
--key / -k text Yes — License key (NM-PRO-XXXX-XXXX-XXXX)

nmem show

Show full content + metadata for a memory by ID.

nmem show [OPTIONS]
Option Type Required Default Description
memory_id text Yes — (positional argument)
--shared / -S boolean No False Use shared/remote storage
--json / -j boolean No False Output as JSON

Works for typed fibers, untyped fibers, and bare neurons. Useful for debugging when recall returns an ID that other commands can't resolve.

nmem storage status

Show current storage backend status.

nmem storage status [OPTIONS]

Displays which backend is active, Pro license status, and whether data files exist for each backend.

nmem storage switch

Switch active storage backend.

nmem storage switch [OPTIONS]
Option Type Required Default Description
backend text Yes — (positional argument)

Switches config.toml to use the specified backend. Data must already exist for the target backend (run migrate first).

nmem sync activate

Activate a NeuralMemory Pro license key.

nmem sync activate [OPTIONS]
Option Type Required Default Description
--key / -k text Yes — License key (NM-PRO-XXXX-XXXX-XXXX)

nmem sync disable

Disable shared mode and use local storage.

nmem sync disable [OPTIONS]

nmem sync enable

Enable shared mode to connect to a remote NeuralMemory server.

nmem sync enable [OPTIONS]
Option Type Required Default Description
server_url text Yes — (positional argument)
--api-key / -k text No — API key for authentication
--timeout / -t float No 30.0 Request timeout in seconds

When shared mode is enabled, all memory operations (remember, recall, etc.) will use the remote server instead of local storage.

nmem sync status

Show shared mode status and configuration.

nmem sync status [OPTIONS]
Option Type Required Default Description
--json / -j boolean No False Output as JSON

nmem sync sync

Manually sync local brain with remote server.

nmem sync sync [OPTIONS]
Option Type Required Default Description
--direction / -d text No both Sync direction: push, pull, or both
--json / -j boolean No False Output as JSON

Directions: push - Upload local brain to server pull - Download brain from server to local both - Full bidirectional sync (default)

nmem sync test

Test connection to the shared server.

nmem sync test [OPTIONS]

nmem unpin

Unpin a fiber (removes grounding; allows decay/prune again).

nmem unpin [OPTIONS]
Option Type Required Default Description
fiber_id text Yes — (positional argument)
--shared / -S boolean No False Use shared/remote storage
--json / -j boolean No False Output as JSON

nmem watch

Watch a directory and auto-ingest files into memory.

nmem watch [OPTIONS]
Option Type Required Default Description
directory text No `` (positional argument)
--action / -a text No scan Action: scan, start, stop, status
--json / -j boolean No False Output as JSON

Auto-generated by scripts/gen_cli_docs.py from Typer app introspection — 82 commands.