Reference

CLI Reference

Complete command reference for the mx CLI.

Full command reference for @memnexus-ai/cli. Install with:

npm install -g @memnexus-ai/cli

Global options

Option	Description
`--no-color`	Disable colored output
`--version`	Show CLI version
`--help`	Show help for any command

auth

Manage authentication and API keys

`mx auth login`

Configure API key for authentication

Option	Description	Default
`--api-key <key>`	API key to store	—
`--interactive`	Prompt for API key (mask input)	—

`mx auth logout`

Remove stored credentials

`mx auth status`

Show current authentication status

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

config

Manage CLI configuration

`mx config get <key>`

Get a configuration value

`mx config set <key> <value>`

Set a configuration value

`mx config list`

List all configuration values

`mx config reset`

Reset configuration to defaults

`mx config agent`

Show AI agent detection status and effective defaults

Option	Description	Default
`--format <format>`	Output format (json\|table)	`table`

`mx config env`

Show all supported environment variables and their current values

`mx config dev`

Switch all URLs to the dev environment

`mx config prod`

Switch all URLs to the production environment

memories

Manage memories

`mx memories list`

List memories

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--last <number>`	Get N most recent memories (shortcut for --limit N --sort createdAt:desc)	—
`--sort <field>`	Sort by field (createdAt:asc, createdAt:desc, eventTime:asc, eventTime:desc)	—
`--named`	List only named memories	—
`--recent <duration>`	Filter: memories from last duration (e.g., 7d, 2w, 1h, 30m)	—
`--after <date>`	Filter: memories created after this date (ISO 8601)	—
`--before <date>`	Filter: memories created before this date (ISO 8601)	—
`--topics <topics>`	Filter: comma-separated list of topics (matches ANY)	—
`--topic <topics>`	Alias for --topics	—
`--exclude-topics <topics>`	Filter: exclude memories with these topics (comma-separated)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories get [id] [moreIds]`

Get one or more memories by ID (supports piping with --stdin)

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—
`--name <name>`	Get memory by name instead of ID	—
`--detail <level>`	Detail level: minimal, standard (default), full	`standard`
`--similar`	Find semantically similar memories (single ID only)	—
`--conversation`	Find memories from the same conversation (single ID only)	—
`--related`	Find memories with shared topics (single ID only)	—
`--limit <number>`	Maximum number of related results	`5`
`--stdin`	Read memory IDs from stdin (one per line or space-separated)	—
`--include-vectors`	Include embedding vectors in json/yaml output (debugging; stripped by default)	—

`mx memories create`

Create a new memory

Option	Description	Default
`--conversation-id <id>`	Conversation ID (REQUIRED - use "NEW" to create a new conversation)	—
`--content <text>`	Memory content	—
`--name <name>`	Assign a kebab-case name for deterministic retrieval	—
`--memory-type <type>`	Memory type (episodic\|semantic\|procedural)	—
`--context <context>`	Context identifier	—
`--topics <topics>`	Comma-separated topics	—
`--role <role>`	Role (user\|assistant\|system)	—
`--importance <number>`	Importance score (0-1)	—
`--supersedes <id>`	ID of memory this supersedes (auto-creates SUPERSEDES relationship)	—
`--code-context <json>`	Structured context tag (JSON object). Example: '{"product":"memnexus","team":"mcp","role":"team-lead"}'. Scopes knowledge to a product/team/service/role for retrieval.	—
`--product <name>`	Shorthand for codeContext.product	—
`--repo <name>`	Shorthand for codeContext.repo	—
`--service <name>`	Shorthand for codeContext.service	—
`--component <name>`	Shorthand for codeContext.component	—
`--team <name>`	Shorthand for codeContext.team	—
`--code-role <name>`	Shorthand for codeContext.role (authorship role, e.g. team-lead)	—
`--task-id <id>`	Task identifier for agentic loop memories (groups attempts at the same problem)	—
`--outcome <outcome>`	Task outcome: resolved \| unresolved	—
`--distill`	Request LLM distillation into structured 5-section notebook (Pro/Enterprise only)	—
`--interactive`	Interactive mode	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories update [id]`

Update an existing memory (by ID or --name)

Option	Description	Default
`--name <name>`	Update a named memory (creates new version via SUPERSEDES chain)	—
`--assign-name <name>`	Assign a name to an existing memory (kebab-case, 2-64 chars)	—
`--rename <newName>`	Rename a named memory (use with --name or an ID that has a name)	—
`--content <text>`	Updated content	—
`--memory-type <type>`	Updated memory type	—
`--context <context>`	Updated context	—
`--topics <topics>`	Updated topics (comma-separated)	—
`--importance <number>`	Updated importance (0-1)	—
`--code-context <json>`	Structured context tag (JSON object). Example: '{"product":"memnexus","team":"mcp","role":"team-lead"}'. Scopes knowledge to a product/team/service/role for retrieval.	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories history <name>`

Show version history for a named memory

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—
`--limit <number>`	Maximum versions to show	`20`

`mx memories delete <id>`

Delete a memory

Option	Description	Default
`--force`	Skip confirmation prompt	—

`mx memories search`

Search memories (keyword, semantic, hybrid; see also: graphrag query)

Option	Description	Default
`--query <text>`	Search query	—
`--limit <number>`	Maximum results	`10`
`--mode <mode>`	Search mode (unified\|content\|facts)	—
`--method <method>`	Search method: keyword (TF-IDF), semantic (vector), hybrid (both)	—
`--threshold <number>`	Minimum score threshold for semantic search (0-1)	`0.5`
`--vector-weight <number>`	Weight for vector search in hybrid mode (0-1)	`0.7`
`--fulltext-weight <number>`	Weight for fulltext search in hybrid mode (0-1)	`0.3`
`--event-from <datetime>`	Filter: event time start (ISO 8601)	—
`--after <datetime>`	Alias for --event-from: event time start (ISO 8601)	—
`--event-to <datetime>`	Filter: event time end (ISO 8601)	—
`--before <datetime>`	Alias for --event-to: event time end (ISO 8601)	—
`--recent <duration>`	Filter: memories from last duration (e.g., 7d, 2w, 1h, 30m)	—
`--topics <topics>`	Filter: comma-separated list of topics (matches ANY)	—
`--topic <topics>`	Alias for --topics	—
`--exclude-topics <topics>`	Filter: exclude memories with these topics (comma-separated)	—
`--type <type>`	Filter: memory type (episodic\|semantic\|procedural)	—
`--conversation-id <id>`	Filter: conversation ID	—
`--preview <number>`	Content preview length in characters	`100`
`--format <format>`	Output format (json\|table\|yaml\|llm\|compact)	—
`--compact`	Compact output: ID, date, first line (shorthand for --format compact)	—
`--id-only`	Output only memory IDs (newline-separated, for piping)	—
`--explain`	Include search scoring explanation	—
`--brief`	Quick lookup preset: --limit 5 --preview 80 --format json	—
`--timeline`	Timeline mode: sort chronologically with timestamps and state indicators. Use --preview <N> (N>100) to show inline content	—
`--group-by <field>`	Group results by field (conversation). Shows memories organized by conversation.	—
`--expand-entities`	Enable entity graph traversal branch in hybrid search	—
`--expand-topics`	Enable topic graph traversal branch in hybrid search	—
`--expand-facts`	Enable fact graph traversal branch in hybrid search	—
`--include-superseded`	Include superseded/contradicted memories in results	—
`--content-only`	Strip output to minimal fields: {id, createdAt, topics, content} (JSON output only)	—
`--effective-only`	Exclude superseded/contradicted memories (passes includeSuperseded: false to API)	—
`--task-id <id>`	Task-scoped retrieval: own-history first, cross-task fills remainder	—
`--outcome-filter <filter>`	Filter by outcome: resolved \| unresolved \| any	—
`--sort <field>`	Sort results by field: score (default), createdAt, updatedAt, eventTime	—
`--order <dir>`	Sort direction: asc \| desc (default: desc)	—
`--scope <pairs>`	Filter by codeContext scope (e.g. repo:memnexus,service:core-api)	—
`--strict`	High-precision mode: fulltext-only, no semantic blending. Use for status/incident/deployment queries where exact-token match matters (sets --mode content --method keyword)	—

`mx memories relationships`

Manage memory relationships (SUPERSEDES, FOLLOWS, RESOLVES, etc.)

`mx memories relationships list <id>`

List relationships for a memory

Option	Description	Default
`--direction <dir>`	Direction: outgoing (from this memory), incoming (to this memory), both	`both`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories relationships create <id>`

Create a relationship from this memory to another

Option	Description	Default
`--target <id>`	Target memory ID	—
`--type <type>`	Relationship type (SUPERSEDES\|FOLLOWS\|RESOLVES\|CONTRADICTS\|REFERENCES)	—
`--confidence <number>`	Confidence score (0-1)	`1.0`
`--reason <text>`	Explanation for the relationship	—
`--interactive`	Interactive mode	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories relationships delete <id> <relationshipId>`

Delete a relationship

Option	Description	Default
`--force`	Skip confirmation prompt	—

`mx memories timeline <id>`

Get timeline context for a memory (related memories before and after)

Option	Description	Default
`--before <number>`	Number of memories before	`5`
`--after <number>`	Number of memories after	`5`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories detect-relationships <id>`

Auto-detect potential relationships for a memory

Option	Description	Default
`--auto-create`	Automatically create high-confidence relationships	—
`--min-confidence <number>`	Minimum confidence threshold (0-1)	`0.7`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories digest`

AI-synthesized summary for humans — reports, handoffs, Slack updates (alias: recall | see also: recap)

Option	Description	Default
`--query <text>`	Search query for the digest (required)	—
`--recent <duration>`	Time window (e.g., 24h, 7d, 2w)	—
`--topics <topics>`	Filter by topics (comma-separated)	—
`--digest-format <format>`	Digest format (structured\|narrative\|timeline\|status-report)	`structured`
`--max-sources <number>`	Maximum source memories to consider	`100`
`--show-sources`	Display source memory IDs	—
`--format <format>`	Output format (json\|yaml\|llm)	—

`mx memories status`

Project status: Done / In Progress / Remaining — AI synthesis of memory corpus (alias for: digest --digest-format status-report)

Option	Description	Default
`--query <text>`	Topic or project to get status for (required)	—
`--recent <duration>`	Time window (e.g., 24h, 7d, 2w)	—
`--topics <topics>`	Filter by topics (comma-separated)	—
`--max-sources <number>`	Maximum source memories to consider	`100`
`--show-sources`	Display source memory IDs	—
`--format <format>`	Output format (json\|yaml\|llm)	—

`mx memories recap`

Quick raw recap of recent work — fast, agent-friendly (see also: digest / recall)

Option	Description	Default
`--recent <duration>`	Time window (e.g., 24h, 7d, 2w)	`24h`
`--query <text>`	Optional search query to filter results	—
`--limit <number>`	Maximum memories to fetch	`200`
`--exclude-topics <topics>`	Exclude memories with these topics (comma-separated)	`auto-saved`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories build-context`

Get contextual briefing before starting work

Option	Description	Default
`--context <text>`	What you are about to work on	—
`--files <paths...>`	File paths you will be touching	—
`--recent <hours>`	How far back to look in hours (default: 24)	`24`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx memories stats`

Aggregated memory analytics (group by day, week, month, or topic)

Option	Description	Default
`--group-by <dimension>`	Group by: day, week, month, topic (required)	—
`--after <date>`	Start date filter (ISO 8601 or YYYY-MM-DD)	—
`--before <date>`	End date filter	—
`--recent <duration>`	Time window (e.g. 1w, 2w, 1mo)	—
`--topic <topic>`	Filter to a specific topic before grouping	—
`--limit <n>`	Max groups to return	`20`
`--format <fmt>`	Output format: table (default), json, csv	—

conversations

Query and manage conversations

`mx conversations list`

List conversations

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--recent <duration>`	Time window (e.g., 24h, 7d, 2w)	—
`--min-memories <n>`	Only show conversations with at least N memories (filters empty init conversations)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx conversations summary <id>`

Get conversation summary

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx conversations timeline <id>`

Get conversation timeline

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx conversations search`

Search conversations

Option	Description	Default
`--query <text>`	Search query	—
`--limit <number>`	Maximum results	`10`
`--format <format>`	Output format (json\|table\|yaml)	—
`--brief`	Quick lookup preset: --limit 5 --format json	—

`mx conversations by-topic`

Find conversations by topic

Option	Description	Default
`--topic <id>`	Topic ID	—
`--topic-id <id>`	Topic ID (alias for --topic)	—
`--format <format>`	Output format (json\|table\|yaml)	—

facts

Manage semantic facts

`mx facts list`

List facts

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx facts get <id>`

Get a specific fact

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx facts create`

Create a new fact

Option	Description	Default
`--subject <text>`	Fact subject (entity name)	—
`--predicate <text>`	Relationship type	—
`--object <text>`	Fact object (related entity)	—
`--confidence <number>`	Confidence score (0-1)	`1.0`
`--memory-id <id>`	Associated memory ID	—
`--interactive`	Interactive mode	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx facts update <id>`

Update an existing fact

Option	Description	Default
`--subject <text>`	Updated subject	—
`--predicate <text>`	Updated predicate	—
`--object <text>`	Updated object	—
`--confidence <number>`	Updated confidence (0-1)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx facts delete <id>`

Delete a fact

Option	Description	Default
`--force`	Skip confirmation prompt	—

`mx facts search`

Search facts

Option	Description	Default
`--query <text>`	Search query	—
`--limit <number>`	Maximum results	`10`
`--format <format>`	Output format (json\|table\|yaml)	—
`--brief`	Quick lookup preset: --limit 5 --format json	—

topics

Manage topics and graph operations

`mx topics list`

List topics

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--context-id <id>`	Filter by context	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics get <id>`

Get topic details

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics merge`

Merge two topics

Option	Description	Default
`--source-id <id>`	Source topic ID	—
`--target-id <id>`	Target topic ID	—
`--force`	Skip confirmation	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics discover-related`

Discover related topics via graph traversal (see also: search)

Option	Description	Default
`--topic <id>`	Topic ID (required)	—
`--topic-id <id>`	Topic ID (alias for --topic)	—
`--context-id <id>`	Filter by context	—
`--limit <number>`	Max results	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics similarity`

Calculate similarity between topics

Option	Description	Default
`--topic-id-1 <id>`	First topic ID (required)	—
`--topic-id-2 <id>`	Second topic ID (required)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics find-similar`

Find similar topics

Option	Description	Default
`--topic-id <id>`	Topic ID (required)	—
`--threshold <number>`	Similarity threshold (0-1)	`0.7`
`--limit <number>`	Max results	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics cluster`

Cluster topics using graph algorithms

Option	Description	Default
`--algorithm <type>`	Clustering algorithm (kmeans\|hierarchical\|dbscan)	`kmeans`
`--num-clusters <number>`	Number of clusters	—
`--context-id <id>`	Filter by context	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics detect-communities`

Detect communities in topic graph

Option	Description	Default
`--algorithm <type>`	Community detection algorithm (louvain\|label-propagation)	`louvain`
`--context-id <id>`	Filter by context	—
`--min-size <number>`	Minimum community size	`2`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx topics search`

Search topics by query string (see also: discover-related)

Option	Description	Default
`--query <text>`	Search query (required)	—
`--limit <number>`	Maximum results	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

entities

Manage entities and graph health

`mx entities list`

List entities

Option	Description	Default
`--type <type>`	Filter by entity type (e.g., PERSON, TECHNOLOGY)	—
`--limit <number>`	Results per page	`10`
`--offset <number>`	Number of entities to skip	`0`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx entities get <id>`

Get entity details

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx entities memories <id>`

Get memories mentioning an entity

Option	Description	Default
`--limit <number>`	Results per page	`10`
`--offset <number>`	Number of memories to skip	`0`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx entities health`

Show graph health and extraction coverage

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

patterns

Manage behavioral patterns

`mx patterns list`

List patterns

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--type <type>`	Pattern type (behavioral\|temporal\|semantic)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx patterns compile`

Compile patterns from memories

Option	Description	Default
`--force`	Force recompilation	—
`--pattern-types <types>`	Comma-separated pattern types to compile	—

`mx patterns update <id>`

Update a pattern

Option	Description	Default
`--description <text>`	Updated description	—
`--active <boolean>`	Enable/disable pattern	—
`--metadata <json>`	Updated metadata	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx patterns feedback`

Record pattern feedback

Option	Description	Default
`--pattern-id <id>`	Pattern ID (required)	—
`--feedback-type <type>`	Feedback type (positive\|negative\|neutral) (required)	—
`--comment <text>`	Feedback comment	—

`mx patterns behavior-state`

Get behavioral state

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx patterns set-behavior-state`

Set behavioral state

Option	Description	Default
`--state <json>`	Behavioral state JSON (required)	—
`--merge`	Merge with existing state instead of replacing	—

`mx patterns detect`

Detect recurring behavioral patterns from memories

Option	Description	Default
`--context-filter <context>`	Filter patterns by context	—
`--timeframe-start <date>`	Start of detection timeframe (ISO 8601)	—
`--timeframe-end <date>`	End of detection timeframe (ISO 8601)	—
`--min-confidence <number>`	Minimum confidence threshold (0-1)	`0.5`
`--max-results <number>`	Maximum patterns to return	`20`
`--auto-store`	Automatically store detected patterns	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx patterns analyze`

Analyze pattern trends, correlations, and generate insights

Option	Description	Default
`--time-range <days>`	Analysis time range in days	`30`
`--group-by <criteria>`	Group analysis by criteria (type\|context\|time)	—
`--include-details`	Include detailed breakdown	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx patterns compile-instructions`

Compile stored behavioral patterns into actionable AI context instructions

Option	Description	Default
`--context <tags>`	Comma-separated topic tags matching current conversation topics (boosts patterns with matching metadata)	—
`--token-budget <n>`	Maximum tokens to use for instructions (50-2000)	`200`
`--min-confidence <n>`	Minimum pattern confidence threshold (0-1)	`0.6`
`--max-instructions <n>`	Maximum number of instructions to return (1-10)	`5`
`--format <format>`	Output format (json\|table)	`table`

graphrag

Execute GraphRAG queries

`mx graphrag query`

Execute GraphRAG query (see also: memories search)

Option	Description	Default
`--query <text>`	Query text (required)	—
`--context-id <id>`	Context ID	—
`--mode <mode>`	Query mode (local\|global\|hybrid)	`hybrid`
`--max-depth <number>`	Graph traversal depth	`3`
`--limit <number>`	Max results	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx graphrag explain <result-id>`

Explain query result

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx graphrag query-communities`

Query communities

Option	Description	Default
`--query <text>`	Query text (required)	—
`--context-id <id>`	Context ID	—
`--limit <number>`	Max communities	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

artifacts

Manage artifacts

`mx artifacts list`

List artifacts

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--context-id <id>`	Filter by context	—
`--type <type>`	Filter by artifact type	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx artifacts get <id>`

Get artifact details

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx artifacts create`

Create artifact

Option	Description	Default
`--name <name>`	Artifact name	—
`--type <type>`	Artifact type	—
`--context-id <id>`	Context ID	—
`--data <json>`	Artifact data as JSON	—
`--metadata <json>`	Artifact metadata as JSON	—
`--interactive`	Interactive mode	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx artifacts update <id>`

Update artifact

Option	Description	Default
`--name <name>`	Updated name	—
`--data <json>`	Updated data	—
`--metadata <json>`	Updated metadata	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx artifacts delete <id>`

Delete artifact

Option	Description	Default
`--force`	Skip confirmation	—

apikeys

Manage API keys

`mx apikeys list`

List API keys

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--format <format>`	Output format (json\|table\|yaml)	—

`mx apikeys create`

Create API key

Option	Description	Default
`--label <label>`	Key label (optional)	—
`--expires-at <date>`	Expiration date (ISO 8601 format, optional)	—

`mx apikeys delete <id>`

Delete API key

Option	Description	Default
`--force`	Skip confirmation	—

system

System health and statistics

`mx system health`

Check system health

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx system status`

Get system status

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx system features`

Get feature flags

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

update

Check for and install CLI updates

`mx update`

Check for and install CLI updates

Option	Description	Default
`--check`	Check for updates without installing	—
`--version <version>`	Install a specific version	—
`--yes`	Skip confirmation prompt	—

narratives

Manage narrative threads (group related memories into storylines)

`mx narratives list`

List narrative threads

Option	Description	Default
`--page <number>`	Page number	`0`
`--limit <number>`	Results per page	`10`
`--state <state>`	Filter by state (open\|resolved\|reopened\|superseded)	—
`--topics <topics>`	Filter by topics (comma-separated)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx narratives get <id>`

Get a specific narrative

Option	Description	Default
`--format <format>`	Output format (json\|table\|yaml)	—

`mx narratives create`

Create a new narrative thread

Option	Description	Default
`--title <text>`	Narrative title	—
`--root-memory-id <id>`	ID of the first memory in this narrative	—
`--topics <topics>`	Comma-separated topics	—
`--interactive`	Interactive mode	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx narratives update <id>`

Update a narrative

Option	Description	Default
`--title <text>`	Updated title	—
`--state <state>`	Updated state (open\|resolved\|reopened\|superseded)	—
`--topics <topics>`	Updated topics (comma-separated)	—
`--format <format>`	Output format (json\|table\|yaml)	—

`mx narratives delete <id>`

Delete a narrative (memories are preserved)

Option	Description	Default
`--force`	Skip confirmation prompt	—

`mx narratives timeline <id>`

Get timeline of memories in a narrative

Option	Description	Default
`--limit <number>`	Maximum memories to return	`20`
`--page <number>`	Page number	`0`
`--format <format>`	Output format (json\|table\|yaml)	—

setup

Configure MCP server integration for AI coding agents

`mx setup verify`

Health-check all configured agents

`mx setup rules`

Write or update MemNexus steering rules for an agent

Option	Description	Default
`--agent <id>`	Agent to target (default: claude-code)	`claude-code`
`--project <dir>`	Project directory (default: global scope)	—
`--dry-run`	Preview changes without writing files	—
`--print`	Print rules to stdout instead of writing to a file (for copy-paste)	—
`--no-markers`	Omit BEGIN/END comment markers from output (with --print)	—

`mx setup remove`

Remove MemNexus config from an agent

Option	Description	Default
`--agent <id>`	Agent to remove config from	—
`--project <dir>`	Project directory (default: global scope)	—

`mx setup hooks`

Install or remove the CommitContext post-commit hook

Option	Description	Default
`--project <dir>`	Project directory (default: cwd)	—
`--remove`	Remove the hook instead of installing	—
`--yes`	Skip confirmation prompts	—
`--dry-run`	Preview changes without writing files	—

`mx setup skills`

Install or manage MemNexus Claude Code slash commands (/mx-save, /mx-checkpoint, /mx-buildcontext)

Option	Description	Default
`--dry-run`	Preview changes without writing files	—
`--force`	Overwrite even if versions match	—
`--remove`	Remove all mx-* skills	—
`--list`	Show installed skills and their versions	—

mcp

MCP bridge commands

`mx mcp serve`

Start stdio↔HTTP MCP bridge (used by AI agents)

Option	Description	Default
`--url <url>`	MCP server URL	`https://mcp.memnexus.ai/mcp`
`--verbose`	Log full JSON-RPC message payloads	—
`--inject-rules`	Inject MemNexus steering rules into the MCP initialize response (used for Claude Desktop)	—

`mx mcp logs`

Show MCP bridge log file location and recent output

Option	Description	Default
`-n, --lines <count>`	Number of lines to show	`50`
`-f, --follow`	Follow log output (like tail -f)	—
`--path`	Only print the log file path	—

`mx mcp status`

Show MCP bridge connection status

Option	Description	Default
`--json`	Output as JSON	—

`mx mcp install`

Configure MCP bridge for an AI coding agent (alias for mx setup)

Option	Description	Default
`--client <id>`	Agent to configure: claude-code \| claude-desktop \| copilot \| cursor \| codex \| opencode \| chatgpt. Omit for interactive picker.	—
`--project <dir>`	Project directory (default: global scope)	—
`--yes`	Skip confirmation prompts	—
`--dry-run`	Preview changes without writing files	—
`--skip-hooks`	Skip lifecycle hook installation	—

`mx mcp hooks`

Manage MemNexus lifecycle hooks

`mx mcp hooks remove`

Remove MemNexus lifecycle hooks from an AI coding agent

Option	Description	Default
`--client <id>`	Agent to remove hooks from (default: claude-code)	`claude-code`
`--project <dir>`	Project directory (default: global scope)	—

`mx mcp restart`

Restart the MCP bridge (kills running bridge processes so the MCP client auto-restarts them)

`mx mcp doctor`

Run a full MCP health check (bridge status + agent config verification)

Option	Description	Default
`--json`	Output as JSON	—

`mx mcp list`

List AI coding agents and their MemNexus MCP configuration status

Option	Description	Default
`--json`	Output as JSON	—

hook

Git hook handlers

`mx hook post-commit`

Capture commit context (called by post-commit hook)

Option	Description	Default
`--cwd <dir>`	Working directory	`/workspace/.worktrees/marketing-20260313/docs-site`
`--dry-run`	Print output without saving	—
`--no-sync`	Skip API sync (cache only)	—
`--summarizer <type>`	Summarizer: llm (default) or heuristic	`llm`
`--no-enrich`	Skip commit message enrichment	—

commit-context

View captured commit context entries

`mx commit-context list`

List recent commit contexts

Option	Description	Default
`--local`	Read from local cache only (no API)	—
`--recent <period>`	Time filter (e.g. 24h, 7d)	`7d`
`--limit <n>`	Max results	`10`
`--branch <name>`	Filter by branch name	—
`--agent <name>`	Filter by agent name (e.g. claude-code)	—
`--format <format>`	Output format	—

`mx commit-context show <sha>`

Show full commit context for a specific commit

Option	Description	Default
`--local`	Read from local cache only	—
`--format <format>`	Output format	—

`mx commit-context sync`

Upload unsynced local cache entries to the API

Option	Description	Default
`--limit <n>`	Max entries to sync	`50`
`--dry-run`	Show what would be synced without syncing	—
`--format <format>`	Output format	—

Environment variables

Variable	Description	Default
`MX_API_URL`	API base URL	`https://api.memnexus.ai`
`MX_API_KEY`	Authentication key	—
`MX_OUTPUT_FORMAT`	Default output format	`table`
`MX_DEFAULT_PAGE_SIZE`	Pagination size	`20`

Platform Examples

Working examples of MemNexus memory integration for Claude Desktop, Claude Code, and Cursor.

Memories API

Memory management and retrieval endpoints

Global options auth mx auth loginmx auth logoutmx auth statusconfig mx config get <key>mx config set <key> <value>mx config listmx config resetmx config agentmx config envmx config devmx config prodmemories mx memories listmx memories get [id] [moreIds]mx memories createmx memories update [id]mx memories history <name>mx memories delete <id>mx memories searchmx memories relationshipsmx memories relationships list <id>mx memories relationships create <id>mx memories relationships delete <id> <relationshipId>mx memories timeline <id>mx memories detect-relationships <id>mx memories digestmx memories statusmx memories recapmx memories build-contextmx memories statsconversations mx conversations listmx conversations summary <id>mx conversations timeline <id>mx conversations searchmx conversations by-topicfacts mx facts listmx facts get <id>mx facts createmx facts update <id>mx facts delete <id>mx facts searchtopics mx topics listmx topics get <id>mx topics mergemx topics discover-relatedmx topics similaritymx topics find-similarmx topics clustermx topics detect-communitiesmx topics searchentities mx entities listmx entities get <id>mx entities memories <id>mx entities healthpatterns mx patterns listmx patterns compilemx patterns update <id>mx patterns feedbackmx patterns behavior-statemx patterns set-behavior-statemx patterns detectmx patterns analyzemx patterns compile-instructionsgraphrag mx graphrag querymx graphrag explain <result-id>mx graphrag query-communitiesartifacts mx artifacts listmx artifacts get <id>mx artifacts createmx artifacts update <id>mx artifacts delete <id>apikeys mx apikeys listmx apikeys createmx apikeys delete <id>system mx system healthmx system statusmx system featuresupdate mx updatenarratives mx narratives listmx narratives get <id>mx narratives createmx narratives update <id>mx narratives delete <id>mx narratives timeline <id>setup mx setup verifymx setup rulesmx setup removemx setup hooksmx setup skillsmcp mx mcp servemx mcp logsmx mcp statusmx mcp installmx mcp hooksmx mcp hooks removemx mcp restartmx mcp doctormx mcp listhook mx hook post-commitcommit-context mx commit-context listmx commit-context show <sha>mx commit-context syncEnvironment variables