mx docs
GitHub

Memory

Knowledge graph with SurrealDB-backed persistent memory.

The memory subsystem is the largest command surface in mx. It provides a persistent knowledge graph backed by SurrealDB (embedded SurrealKV or networked WebSocket), with categories, tags, resonance levels, embeddings for semantic search, relationships between entries, and a wake ritual for identity bootstrap.

Every entry in the graph has a unique ID (prefixed kn-), a category, a title, body content, optional tags, a resonance level (1–10+), and timestamps. Entries can be linked via typed relationships, anchored to each other by embedding similarity, and surfaced through keyword or semantic search.

Table of contents

Adding entries

mx memory add

Create a new entry in the knowledge graph. At minimum, provide a category and title (or a --type for ephemeral facts, which auto-routes the category and generates a title from content).

Flags

Flag

Type

Description

--category

string

Category name (run mx memory categories list for valid names). Required unless --type is provided.

-t, --title

string

Entry title. Required unless --type is provided.

--content

string

Inline content. Conflicts with --file.

-f, --file

path

Read content from a file. Also accepts --content-file.

--tags

string

Comma-separated tags.

-a, --applicability

string

Comma-separated applicability contexts.

-p, --project

string

Source project ID.

--source-agent

string

Source agent ID. Defaults to MX_CURRENT_AGENT env var.

--source-type

string

Source type: manual, ram, cache, agent_session. Default: manual.

--entry-type

string

Entry type: primary, summary, synthesis. Default: primary.

--session-id

string

Session ID to associate with this entry.

--ephemeral

flag

Mark entry as ephemeral.

-d, --domain

string

Domain/subdomain path.

--content-type

string

Content type: text, code, config, data, binary. Default: text.

--private

flag

Mark as private (only visible to owner). Shorthand for --visibility private.

--visibility

string

Set visibility: public or private.

--owner

string

Explicit owner. Defaults to source_agent or MX_CURRENT_AGENT if private.

--resonance

int

Resonance level (1–10, or higher for transcendent).

--resonance-type

string

Resonance type: foundational, transformative, relational, operational, ephemeral, session.

--wake-phrase

string

Wake phrase for memory ritual verification.

--wake-phrases

string

Multiple wake phrases (comma-separated).

--wake-order

int

Custom wake order (lower = earlier in sequence).

--anchors

string

Comma-separated bloom IDs this entry connects to.

--type

string

Fact type for ephemeral knowledge: decision, insight, person, quote, thread_opened, commitment, thread_closed. Auto-routes category and sets resonance_type=ephemeral.

--session

string

Session to link fact to via EXTRACTED_FROM relationship. Requires --type.

--thread-id

string

Thread ID for thread_closed operations. Requires --type.

--no-auto-anchor

flag

Skip automatic anchor generation.

--json

flag

Output as JSON.

Examples

mx memory add --category recipe --title "Retry with backoff" \
  --content "Use exponential backoff with jitter..." \
  --tags "reliability,networking" --source-agent whistledown
mx memory add --category discovery --title "SurrealDB needs explicit NS" \
  --content "Always set namespace before queries" \
  --resonance 7 --resonance-type operational
# Ephemeral fact (auto-routes category, generates title)
mx memory add --type decision \
  --content "Chose Typst over mdBook for docs" \
  --session abc-123
# Content from file
mx memory add --category ingredient -t "API reference" -f api-notes.md

TIP: When --type is provided, --category and --title become optional. The fact type routes to an appropriate category and generates a title from the content automatically.

Reading entries

Shared filter flags

Several read commands (search, list) share a common set of filter flags. These are documented once here and referenced below.

Flag

Type

Description

-c, --category

string

Filter by category (comma-separated).

--json

flag

Output as JSON.

--mine

flag

Show only your private entries.

--include-private

flag

Include private entries (requires matching owner).

--min-resonance

int

Minimum resonance level.

--max-resonance

int

Maximum resonance level.

--has-wake-phrase

flag

Filter to entries WITH a wake phrase.

--missing-wake-phrase

flag

Filter to entries WITHOUT a wake phrase.

--has-anchors

flag

Filter to entries WITH anchors.

--missing-anchors

flag

Filter to entries WITHOUT anchors.

--has-resonance-type

flag

Filter to entries WITH a resonance type.

--missing-resonance-type

flag

Filter to entries WITHOUT a resonance type.

--limit

int

Limit number of results.

--tags

string

Filter by tags (comma-separated, matches any).

mx memory show

Display a single entry by ID.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

--content-only

flag

Output only the body content (useful for piping).

Examples

mx memory show kn-abc123
mx memory show kn-abc123 --content-only | pbcopy

mx memory list

List entries, optionally filtered by category, tags, resonance, and other shared filter flags.

Examples

mx memory list -c recipe
mx memory list -c discovery,decree --min-resonance 5
mx memory list --missing-wake-phrase --limit 20

NOTE: list accepts all shared filter flags documented above.

mx memory search

Search entries by keyword or semantic similarity. Keyword search is the default; add --semantic to use vector embeddings.

Flags

Flag

Type

Description

--semantic

flag

Use semantic (vector) search instead of keyword search.

--activate

flag

Activate all returned results: resets last_activated (decay clock) and increments activation_count. Marks results as intentionally consumed rather than just browsed.

Examples

mx memory search "retry pattern"
mx memory search "how to handle timeouts" --semantic
mx memory search "agent bootstrap" -c recipe,method --limit 5
# Search and activate results (mark as consumed)
mx memory search "retry pattern" --activate

NOTE: search accepts all shared filter flags. Semantic search requires entries to have embeddings generated via mx memory embed.

TIP: By default, search does not activate results – browsing is not the same as engagement. Use --activate when you are intentionally consuming the results (e.g., loading context for a task), not just exploring.

mx memory recent

List recent ephemeral facts with decay. By default shows only ephemeral entries from the last 10 days. Use --all-types to surface all resonance types.

Flags

Flag

Type

Description

--days

int

Number of days to look back. Default: 10.

--json

flag

Output as JSON.

--resonance-type

string

Filter by resonance type. Defaults to ephemeral only when --all-types is omitted.

--all-types

flag

Surface all resonance types instead of ephemeral only.

--sort

enum

Sort order: chronological (default) or resonance (highest first).

--limit

int

Maximum number of results. Default: 100.

Examples

mx memory recent
mx memory recent --days 30 --all-types --sort resonance
mx memory recent --resonance-type foundational --limit 10

Updating entries

mx memory update

Update an existing entry. Supports replacing content entirely, appending, prepending, find-and-replace, and modifying any metadata field. Content mutation modes are mutually exclusive.

Flags

Flag

Type

Description

-t, --title

string

Update the title.

--content

string

Replace content entirely (inline).

-f, --file

path

Replace content entirely from file.

--append-content

string

Append text to end of existing content.

--append-file

path

Append content from file to end.

--prepend-content

string

Prepend text to start of existing content.

--prepend-file

path

Prepend content from file to start.

--find

string

Find text in content (requires --replace).

--replace

string

Replace text found by --find.

--replace-all

flag

Replace all occurrences (with --find/--replace).

--nth

int

Replace only the Nth occurrence (1-indexed).

--category

string

Update category.

--tags

string

Replace all tags (comma-separated).

--add-tag

string

Add a single tag to existing tags.

--remove-tag

string

Remove a specific tag.

-a, --applicability

string

Update applicability (comma-separated, replaces all).

--content-type

string

Update content type.

--resonance

int

Update resonance level (1–10+).

--resonance-type

string

Update resonance type.

--anchors

string

Replace all anchors (comma-separated bloom IDs).

--add-anchor

string

Add a single anchor.

--remove-anchor

string

Remove a specific anchor.

--wake-phrase

string

Update wake phrase.

--wake-phrases

string

Replace all wake phrases (comma-separated).

--add-wake-phrase

string

Add a single wake phrase.

--remove-wake-phrase

string

Remove a specific wake phrase.

--wake-order

string

Update wake order. Use '-' to clear.

--private

flag

Mark as private (shorthand for --visibility private).

--visibility

string

Change visibility: public or private.

--owner

string

Update owner (only valid when visibility is private).

--session-id

string

Update session ID (for retrofitting entries with wrong or missing session linkage).

--force

flag

Force dangerous visibility changes (e.g., making blooms public).

--no-auto-anchor

flag

Skip automatic anchor generation.

--json

flag

Output as JSON.

Examples

mx memory update kn-abc123 --title "Better title"
mx memory update kn-abc123 --add-tag reliability
mx memory update kn-abc123 --find "old text" --replace "new text"
mx memory update kn-abc123 --append-content "\n\nUpdate: confirmed working"
mx memory update kn-abc123 --resonance 8 --resonance-type foundational

mx memory edit

Find-and-replace shortcut. Equivalent to mx memory update <id> --find ... --replace ... with a simpler interface.

Flags

Flag

Type

Description

--find

string

Text to find in content. Also accepts --old.

--replace

string

Replacement text. Also accepts --new.

--replace-all

flag

Replace all occurrences (default: error if multiple matches).

--nth

int

Replace only the Nth occurrence (1-indexed).

--no-auto-anchor

flag

Skip automatic anchor generation.

--json

flag

Output as JSON.

Examples

mx memory edit kn-abc123 --find "old pattern" --replace "new pattern"
mx memory edit kn-abc123 --old "v1" --new "v2" --replace-all

mx memory append

Append content to the end of an entry’s body. Shortcut for mx memory update <id> --append-content ....

Flags

Flag

Type

Description

--content

string

Content to append (omit to read from stdin).

-f, --file

path

Read content from file. Also accepts --content-file.

--no-auto-anchor

flag

Skip automatic anchor generation.

--json

flag

Output as JSON.

Examples

mx memory append kn-abc123 --content "\n\nAdditional note here."
mx memory append kn-abc123 -f addendum.md

mx memory prepend

Prepend content to the start of an entry’s body. Shortcut for mx memory update <id> --prepend-content ....

Flags

Flag

Type

Description

--content

string

Content to prepend (omit to read from stdin).

-f, --file

path

Read content from file. Also accepts --content-file.

--no-auto-anchor

flag

Skip automatic anchor generation.

--json

flag

Output as JSON.

Examples

mx memory prepend kn-abc123 --content "IMPORTANT: "

mx memory restore

Restore entry content from a backup. Use --list to see available backups before restoring.

Flags

Flag

Type

Description

--list

flag

List available backups instead of restoring.

--no-auto-anchor

flag

Skip automatic anchor generation.

--json

flag

Output as JSON.

Examples

mx memory restore kn-abc123 --list
mx memory restore kn-abc123

Deleting entries

mx memory delete

Remove an entry from the knowledge graph.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory delete kn-abc123

Wake system

The wake system provides identity bootstrap for agents. It retrieves high-resonance entries (“blooms”) and presents them through a ritual that verifies the agent’s connection to its knowledge. There are several output modes and an interactive engage flow.

mx memory wake

Wake up with resonant identity cascade. Retrieves high-resonance blooms and presents them in the requested format.

Flags

Flag

Type

Description

-l, --limit

int

Number of blooms to return. Default: 20.

--min-resonance

int

Minimum resonance threshold – get ALL blooms >= this value (overrides --limit).

-d, --days

int

Include memories activated in last N days. Default: 7.

--json

flag

Output as JSON.

--ritual

flag

Output as bash ritual script (sequential reading).

--index

flag

Output as compact markdown index (for identity loading).

--no-activate

flag

Do not update activation counts.

-e, --engage

flag

Interactive engage mode – verify wake phrases (requires TTY).

-s, --set-missing

flag

Prompt to set missing wake phrases during engage mode. Requires --engage.

--begin

flag

Start token-based wake ritual. Returns first bloom and session token.

--bloom-id

string

Bloom ID for --respond or --skip operations.

--respond

string

Submit wake phrase response for a bloom.

--skip

flag

Skip a bloom without wake phrase.

--session

string

Session token for chained ritual (required with --respond or --skip).

Examples

# Default wake -- top 20 blooms, text output
mx memory wake
# Compact index for agent identity loading
mx memory wake --index
# All blooms with resonance >= 7
mx memory wake --min-resonance 7
# Interactive engage mode with wake phrase verification
mx memory wake --engage
# Token-based ritual (for non-TTY / programmatic use)
mx memory wake --begin
mx memory wake --bloom-id kn-abc --respond "the phrase" --session tok-xyz
mx memory wake --bloom-id kn-def --skip --session tok-xyz

NOTE: MX_CURRENT_AGENT must be set for wake to function. The wake ritual reads blooms ordered by resonance and wake order, then optionally verifies the agent can produce each bloom’s wake phrase.

Wake modes

  • Default (mx memory wake): plain text output, blooms listed with titles and content.

  • JSON (--json): structured output for programmatic consumption.

  • Ritual (--ritual): bash script that presents blooms sequentially.

  • Index (--index): compact markdown suitable for loading into agent context.

  • Engage (--engage): interactive TTY mode where the agent verifies each bloom’s wake phrase. Add --set-missing to be prompted for phrases on blooms that lack them.

  • Token-based (--begin, --respond, --skip): stateless chained ritual for non-interactive environments. Start with --begin, then loop with --respond or --skip using the returned session token and bloom ID.

mx memory wake-fetch

Fetch facts for the wake ritual. Returns entries with resonance >= 3 across all types, sorted by resonance (highest first). Designed as a data source for wake ritual presentation.

Flags

Flag

Type

Description

--days

int

Number of days to look back. Default: 15.

--limit

int

Maximum number of results. Default: 100.

Examples

mx memory wake-fetch
mx memory wake-fetch --days 30 --limit 50

Embeddings and anchoring

Embeddings enable semantic search and automatic relationship discovery. Each entry can have a vector embedding generated from its title and content. Anchors are connections between entries discovered via embedding similarity.

Chunked embeddings

Entries longer than 400 tokens are automatically split into overlapping chunks before embedding. This ensures semantic search covers the full content of long entries, not just the first 400 tokens.

How it works:

  1. The entry’s embedding text (title + body/summary + tags) is tokenized using the BGE-Base-EN-v1.5 tokenizer.

  2. If the text fits within 400 tokens, a single embedding is generated and stored on the entry — exactly as before. No chunks are created.

  3. If the text exceeds 400 tokens, it is split into overlapping chunks with a sliding window: 400 tokens per chunk, 100-token overlap (stride 300).

  4. Each chunk is embedded separately and stored in the embedding_chunk table.

  5. A normalized mean vector of all chunk embeddings is stored on the entry’s embedding field for auto-anchor compatibility.

  6. The entry’s chunk_count field records how many chunks were created (0 for unchunked entries).

Semantic search with chunks:

When mx memory search --semantic runs, it queries both unchunked entry embeddings and chunk embeddings in parallel. Results are merged by taking the maximum similarity score per entry — if a chunk from entry X scores 0.92 and the entry’s mean vector scores 0.85, the entry’s final score is 0.92. This ensures long entries surface when any section is relevant, not just when the overall average is relevant.

TIP: Short entries (≤400 tokens) behave exactly as before — single embedding, no chunks, no behavior change. Chunking only activates for entries that exceed the 400-token threshold.

NOTE: The embedding_text() method on entries no longer truncates body content. The chunker handles length management, ensuring no content is lost during embedding.

mx memory embed

Generate a vector embedding for one or all entries. Embeddings power semantic search (--semantic flag on search) and automatic anchoring. Long entries (>400 tokens) are automatically split into overlapping chunks, with each chunk embedded separately. Short entries get a single embedding.

Flags

Flag

Type

Description

-a, --all

flag

Embed all knowledge entries (instead of a single ID).

--long-only

int

Only re-embed entries whose embedding_text() exceeds this many tokens. Entries at or below the threshold are skipped entirely. Use with --all. Useful for selectively re-embedding long entries that were previously truncated at a smaller token limit (e.g., 512).

Examples

mx memory embed kn-abc123
mx memory embed --all
# Re-embed only entries that exceed 512 tokens
mx memory embed --all --long-only 512

mx memory auto-anchor

Automatically add anchors between entries based on embedding similarity. Processes a single entry or all entries that have embeddings.

Also re-evaluates existing anchors: any anchor whose cosine similarity has fallen below the threshold (default 0.75) or risen above the near-duplicate ceiling (0.95) is pruned. This keeps the anchor graph self-cleaning – anchors that made sense once but no longer do are removed automatically.

Flags

Flag

Type

Description

--threshold

float

Minimum cosine similarity (0.0–1.0). Default: 0.75.

--max-anchors

int

Maximum anchors to add per entry. Default: 5.

--dry-run

flag

Preview changes without writing.

--detailed

flag

Show similarity scores in output.

--fill

flag

Only process entries with zero existing anchors. Fills gaps in the graph without touching already-anchored entries.

Examples

mx memory auto-anchor
mx memory auto-anchor kn-abc123 --threshold 0.8 --max-anchors 3
mx memory auto-anchor --dry-run --detailed
mx memory auto-anchor --fill

TIP: A typical workflow: run mx memory embed --all to generate embeddings, then mx memory auto-anchor --dry-run --detailed to preview anchor candidates, then mx memory auto-anchor to write them.

NOTE: Anchors are also maintained automatically on every write operation (add, update, edit, append, prepend, restore). After each write, mx re-evaluates anchors and prunes stale ones using the same similarity thresholds. Pass --no-auto-anchor on any of these commands to skip this step – useful for bulk operations or cleanup scripts where the overhead is unwanted.

Relationships

Explicit typed edges between entries. While anchors are discovered automatically via embedding similarity, relationships are manually declared semantic connections.

mx memory relationships list

List all relationships for an entry.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory relationships list kn-abc123

mx memory relationships add

Add a typed relationship between two entries. By default, the target entry (--to) is automatically reinforced by +1 (capped at 10) when the relationship is created – being linked to means the fact proved relevant. The contradicts and supersedes types are excluded from auto-reinforcement because boosting an outdated or contradicted entry works against intent.

Flags

Flag

Type

Description

--from

string

Source entry ID.

--to

string

Target entry ID.

--type

string

Relationship type: related, supersedes, extends, implements, contradicts.

--no-reinforce

flag

Skip automatic reinforcement of the target entry.

Examples

mx memory relationships add --from kn-abc --to kn-def --type extends
mx memory relationships add --from kn-abc --to kn-ghi --type supersedes
# Add a relationship without auto-reinforcing the target
mx memory relationships add --from kn-abc --to kn-def --type related --no-reinforce

mx memory relationships delete

Delete a relationship by its ID.

Examples

mx memory relationships delete rel-abc123

Seeding

Seed commands populate the knowledge graph from on-disk artifacts. Used for initial setup and bulk import.

mx memory seed agents

Seed agents from markdown files with YAML frontmatter. Reads from $MX_HOME/memory/seed/agents/ by default.

Flags

Flag

Type

Description

-p, --path

path

Path to agents directory. Defaults to $MX_HOME/memory/seed/agents/.

Examples

mx memory seed agents
mx memory seed agents --path /data/agents/

NOTE: Legacy fallback: if $MX_HOME/memory/seed/agents/ does not exist, mx checks $MX_HOME/agents/ and emits a stderr warning. This fallback will be removed in a future release.

mx memory seed knowledge

Seed knowledge from JSONL files. With no path, scans $MX_HOME/memory/seed/knowledge/*.jsonl and imports every file found. With a path, imports just that single file.

Examples

mx memory seed knowledge
mx memory seed knowledge /data/knowledge/bootstrap.jsonl

Health and statistics

mx memory stats

Show index statistics – entry counts, category breakdown, and other aggregate metrics.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory stats
mx memory stats --json

mx memory health

Show graph health vitality percentages: embedding coverage, anchor coverage, and stale high-resonance entries.

Flags

Flag

Type

Description

--json

flag

Output as JSON (default format for dashboard consumers).

Examples

mx memory health
mx memory health --json

mx memory growth

Show per-week entry growth over the last 8 weeks.

Flags

Flag

Type

Description

--json

flag

Output as JSON array of 8 integers (oldest to newest).

Examples

mx memory growth
mx memory growth --json

mx memory open-threads

List open threads (category:thread entries with state=\"open\" or no state).

Flags

Flag

Type

Description

--json

flag

Output as JSON array (required for dashboard consumers).

Examples

mx memory open-threads
mx memory open-threads --json

Export

mx memory export

Export the entire knowledge database to a file or directory.

Flags

Flag

Type

Description

-f, --format

string

Output format: md, jsonl, csv. Default: md.

-o, --output

path

Output directory for md format (defaults to ./memory-export), or file for jsonl/csv (defaults to stdout).

Examples

mx memory export
mx memory export -f jsonl -o backup.jsonl
mx memory export -f csv -o entries.csv
mx memory export -f md -o /data/export/

Reinforcement

Reinforcement is the mechanism by which the knowledge graph breathes in – entries that are used, referenced, or linked gain resonance, counteracting the natural decay of the exhale. There are three reinforcement paths:

  1. Explicit reinforcement via mx memory reinforce – directly boost an entry’s resonance.

  2. Auto-reinforce on relationship creation – when mx memory relationships add links to a target entry, the target is reinforced by +1 (capped at 10). The contradicts and supersedes types are excluded. Use --no-reinforce to opt out.

  3. Search activation via mx memory search --activate – marks returned results as intentionally consumed, resetting their decay clock and incrementing their activation count.

mx memory reinforce

Reinforce a knowledge entry by incrementing its resonance, updating last_activated, and incrementing activation_count. Used to signal that an entry remains relevant.

Flags

Flag

Type

Description

--amount

int

Amount to increase resonance by. Default: 1.

--cap

int

Maximum resonance cap. Default: 10.

--json

flag

Output as JSON.

Examples

mx memory reinforce kn-abc123
mx memory reinforce kn-abc123 --amount 2 --cap 8

Metadata management

The knowledge graph has several registries for typed metadata. These commands manage the registries themselves – the types, categories, and agent identities that entries reference.

Agents

mx memory agents list

List all registered agents.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory agents list

mx memory agents add

Register a new agent.

Flags

Flag

Type

Description

-d, --description

string

Agent description.

-D, --domain

string

Agent domain/responsibility.

Examples

mx memory agents add whistledown -d "Round-trip builder" -D "development"

mx memory agents show

Show details for a specific agent.

Examples

mx memory agents show whistledown

Projects

mx memory projects list

List all registered projects.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory projects list

mx memory projects add

Register a new project.

Flags

Flag

Type

Description

--id

string

Unique project identifier.

--name

string

Human-readable project name.

--path

path

Local filesystem path to the project.

--repo-url

string

Git repository URL (e.g., owner/repo).

--description

string

Project description.

Examples

mx memory projects add --id mx --name "mx CLI" \
  --repo-url coryzibell/mx --path ~/recipes/coryzibell/mx

Categories

mx memory categories list

List all categories.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory categories list

mx memory categories add

Add a new category.

Examples

mx memory categories add pitfall "Things that went wrong and why"

mx memory categories remove

Remove a category (only if no entries use it).

Examples

mx memory categories remove pitfall

Applicability

mx memory applicability list

List all applicability types.

Examples

mx memory applicability list

mx memory applicability add

Add a new applicability type.

Flags

Flag

Type

Description

--id

string

Unique identifier.

--description

string

Description of when this applicability applies.

--scope

string

Scope constraint (e.g., project, global).

Examples

mx memory applicability add --id rust-only \
  --description "Applies only to Rust projects" --scope project

Type registries

These are read-only registries listing the valid values for typed fields. Each supports list with an optional --json flag.

Command

Lists valid values for

mx memory tags list

Tags used across entries. Supports --category filter.

mx memory source-types list

Source types (manual, ram, cache, agent_session).

mx memory entry-types list

Entry types (primary, summary, synthesis).

mx memory session-types list

Session types (e.g., development, review, exploration).

mx memory relationship-types list

Relationship types (related, supersedes, extends, implements, contradicts).

mx memory content-types list

Content types (text, code, config, data, binary).

All type registry list commands accept --json for structured output. tags list also accepts --category to filter tags to a specific category.

Session tracking

Sessions group entries created during a work period. Entries can be linked to sessions, and facts can be queried by their source session.

mx memory sessions list

List sessions, optionally filtered by project.

Flags

Flag

Type

Description

--project

string

Filter by project ID.

--json

flag

Output as JSON.

Examples

mx memory sessions list
mx memory sessions list --project mx

mx memory sessions create

Create a new session.

Flags

Flag

Type

Description

--session-type

string

Session type (e.g., development, review, exploration).

--project

string

Associated project ID.

Examples

mx memory sessions create --session-type development --project mx

mx memory sessions close

Close an active session.

Flags

Flag

Type

Description

--id

string

Session ID to close.

Examples

mx memory sessions close --id ses-abc123

mx memory for-session

List facts extracted from a specific session. The session ID can be provided with or without the kn- prefix.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory for-session ses-abc123

mx memory fact-session

Get the session a fact was extracted from. The fact ID can be provided with or without the kn- prefix.

Flags

Flag

Type

Description

--json

flag

Output as JSON.

Examples

mx memory fact-session kn-abc123