Memory Console guide

User goal and console overview

The Memory Console provides a visual web interface at /me (aliased as /memory-console) for inspecting, searching, and managing memory records, agent connections, projects, and security settings. Rather than interacting solely through CLI commands or MCP tools, developers and team members use the console to audit what agents have recorded, correct inaccurate memories, inspect agent attribution, manage API tokens, and control data export and deletion.

Availability and prerequisites

Memory Console is available in any modern browser for authenticated users. The console layout organizes capabilities into distinct workspace and access sections accessible via sidebar navigation and direct URL hash anchors:

• Authentication requirement: An active user session is required. Unauthenticated visits redirect to /login (the sign-in page).
• Overview (#overview): Quick summary of personal memory workspace, recent activity, connected agents, and storage signals.
• Memories (#memory): Search, filter, inspect, and edit memory records with full-text search, type filters, and source filters.
• Knowledge (#knowledge): View and maintain knowledge documents for authorized agents to consult.
• Projects (#projects): Browse project workspaces, contextual notes, and project-associated memories.
• Cloud Skill (#skills): Author, import, and manage procedural cloud skills. Marked with a Preview badge and available if enabled for your account.
• Dream (#dream): Review recurring patterns and synthesis suggestions before they become durable memory. Marked with a Preview badge and available if enabled for your account.
• Agents (#agents): View agents connected to this workspace, inspect last active timestamps, and review permissions.
• Teams (#teams): Create and manage teams, invite members, and configure team workspace roles. Marked with a Preview badge and available if enabled for your account.
• Profile (#profile): Account identity, email verification status, and personal preferences.
• Sessions (#sessions): Review active sign-in sessions and connected devices.
• API Keys (#api-keys): Issue, rotate, and revoke scoped access tokens for CLI, agents, and automation.
• Settings (#memory-settings): Privacy posture, security modes, data export requests, and memory deletion controls.

One complete task: Inspect and correct synthetic memory

In this walkthrough, you locate the synthetic convention memory saved during the Quickstart guide, inspect its agent attribution, correct its text in the console, verify the update, and confirm personal workspace boundaries.

1. Locate synthetic memory in Memories panel

Navigate to /me and open the Memories panel by clicking Memories in the sidebar or navigating to #memory. Type "demo project timestamp format" into the search bar or filter by the Facts category. The list renders matching memory cards showing summary content, creation time, and visibility badges.

# Access personal workspace
https://xmemo.dev/me#memory

# Access team workspace with explicit team context
https://xmemo.dev/me?space=<team_id>#memory

2. Inspect agent attribution and metadata

Click the memory card to expand the Explain panel. The console displays provenance and attribution details captured at write time:

• Writer kind and agent identity: Displays whether the memory was written by an agent, human user, or imported script, along with agent name and ID.
• Agent instance display: Displays the distinct agent instance identifier (agent_instance_display) to distinguish between separate runs of the same agent.
• Device and transport: Shows the device label (device_label) and transport channel through which the write was submitted.
• Visibility and scope: Indicates whether the record is Personal or Team shared (team_id), along with project and bucket assignments.
• Version history: Displays current revision number and links to the Version lineage drawer.

3. Correct memory content inline

Click the Edit button on the memory card to open the inline editor. Update the content to reflect the refined project convention, then click Save:

• Updated content: Change "demo project timestamp format" to "Demo project timestamps use UTC with ISO 8601 formatting.".
• Save action: The console issues an authenticated update request with CSRF protection, incrementing the revision counter.
• Conflict protection: If an agent updated the memory concurrently, the console detects the revision mismatch (HTTP 409) and prompts with "CAS version conflict: Data has been updated elsewhere. Please refresh and try again." to review the latest content before overwriting.

4. Verify update across console and agent recall

After saving, the console displays a success notification and updates the memory card. The Version lineage drawer shows the previous revision superseded by the new revision. To verify that connected agents immediately observe the correction, execute recall in your terminal:

node scripts/xmemo-skill.mjs recall --query "demo project timestamp format" --limit 1

Observable result

A successful console correction produces observable changes across the web UI and agent interfaces:

• Web console toast: A confirmation toast ("Memory updated") appears at the bottom of the screen.
• Card presentation: The card immediately displays the updated text, refreshed timestamp, and incremented revision index.
• Lineage drawer: The Version lineage drawer displays word-level diffs comparing the prior revision against the updated text.
• Agent recall output: A query for "demo project timestamp format" returns the updated text with its latest revision reference.

What agents can access

Agents and console users operate under distinct authentication and permission boundaries:

• Console permissions: Console users access data through browser cookies with CSRF validation. Actions in the console reflect the logged-in user profile.
• Agent permissions: Agents authenticate using scoped API keys issued in the API Keys panel (#api-keys) or OAuth tokens. Permitted token scopes include memory:read, memory:write, knowledge:read, ledger:read, ledger:write, and read:audit.
• Agent visibility boundary: Agents cannot view account passwords, active browser sessions, or billing details. An agent issued a personal token can only query personal memories and public knowledge; it cannot inspect team workspaces unless specifically bound to that team.

Sharing and storage boundaries

Memory Console enforces strict separation between personal data and team workspaces:

• Space selection: The space selector switches between Personal memory and Team workspaces via the query parameter ?space=<team_id> (managed by useSpaceContext). Omitting the parameter or setting ?space=personal scopes views to personal memory.
• Settings isolation: The Settings panel (#memory-settings) is automatically hidden when browsing a team space to prevent cross-tenant configuration drift. Attempting to open Settings while in a team space resets the active space to Personal.
• Data export boundaries: From Settings (#memory-settings), users can request a secret-safe export archive (/api/v1/me/export-requests). Personal export includes personal memory records, version lineage, personal ledger transactions, and personal documents. Team memory is strictly excluded from personal exports.
• Export download authorization: Generated export archives require a sudo challenge to obtain a temporary download authorization token before retrieving xmemo-personal-export.zip. Export artifacts can be revoked and deleted at any time.
• Memory deletion controls: Users can delete individual memories with confirmation, clear all memories, or reset memories by confirming their account email and typing the confirmation phrase "RESET MY XMEMO MEMORY". Self-service account closure requires transferring any owned teams before deletion (HTTP 409).
• TODO entry points: The console exposes TODOs as aggregate counters within project cards under Projects (#projects), displaying the number of open action items associated with each workspace.
• Ledger entry points: Ledger records appear as a filterable type category under Memories (#memory), can be scoped on API tokens (ledger:read, ledger:write) under API Keys (#api-keys), and are included in personal data export archives.

Failure recovery

Resolve common console errors and access issues using the following steps:

• Memory not found in search: Check if the space selector is set to the correct space (Personal vs Team ?space=<team_id>). Clear active toolbar filter dropdowns (such as Type or Visibility) to broaden search results.
• Concurrent edit error (HTTP 409): If an agent or another tab updated the memory while your edit was open, the console displays "CAS version conflict: Data has been updated elsewhere. Please refresh and try again.". Refresh the list, inspect the new revision, and re-apply your changes.
• Settings option not visible in sidebar: Settings is hidden when viewing a team space. Open the space dropdown in the sidebar and switch back to Personal workspace to restore access to Settings (#memory-settings).
• Export download authorization expired: Sudo download authorization tokens are short-lived. Re-enter your account credentials in the download prompt to generate a fresh download authorization.

Related reference

Consult related concept and tool documentation for underlying mechanisms and API contracts:

• Quickstart: /docs/quickstart for initial setup and synthetic memory creation.
• Memory correction: /docs/concepts/memory-correction for revision history and superseding semantics.
• Memory deletion: /docs/concepts/memory-deletion for soft-deletion and tombstone lifecycle.
• Provenance and attribution: /docs/concepts/provenance-attribution for agent instance and writer kind specifications.
• Projects: /docs/concepts/projects for project workspace boundaries and context scopes.
• Teams: /docs/capabilities/teams for team workspace creation, roles, and collaboration.
• Knowledge bases: /docs/concepts/knowledge-bases for document bases and publication rules.
• Cloud skills: /docs/concepts/cloud-skills for procedural cloud skill authoring.
• Dream reflection: /docs/concepts/dream-reflection for pattern analysis concepts.
• Governance and retention: /docs/concepts/governance-retention for retention policies and DSAR controls.
• Data boundary: /docs/security/data-boundary for encryption, multi-tenancy, and isolation.
• TODO tool: /docs/tools/todos for agent task tracking tool parameters.
• Ledger tool: /docs/tools/ledger for financial transaction tool parameters.