CLI commands

#agentdb workspace list [options]

Workspaces this login can use; * marks the one commands use here

#agentdb workspace create [options] [name]

Create a workspace in your account, e.g. agentdb workspace create "My Application" (needs an account login)

Arguments: [name]

#agentdb workspace use [options] <workspace>

Select a workspace (id, slug or name) for later commands and for agentdb mcp proxy

Arguments: <workspace>

#agentdb workspace current [options]

Show which workspace commands use here, and why

#agentdb workspace rename [options] <workspace> <name>

Rename a workspace, e.g. agentdb workspace rename my-application "New Name" (id and data stay the same)

Arguments: <workspace>; <name>

#agentdb workspace info [options]

Show the current workspace with its quotas

#agentdb workspace delete [options] <workspace>

Not available from the CLI (explorer only since rc.8): the owner deletes a workspace in /app, Settings tab

Arguments: <workspace>

#agentdb workspace restore [options] <workspace>

Not available from the CLI (explorer only since rc.8): the owner restores a workspace in /app, Settings tab

Arguments: <workspace>

#agentdb workspace import [options] <bundle>

Re-import a data-export bundle (.tar.gz from data-export --bundle, or .ndjson) into a NEW workspace of your account (needs an account login)

Arguments: <bundle>

#agentdb schema inspect [options]

List objects, fields and relationships

#agentdb schema object-create [options] <name>

Arguments: <name>

#agentdb schema field-add [options] <object> <name>

Arguments: <object>; <name>

#agentdb schema relationship-add [options] <source_object> <name> <target_object>

Arguments: <source_object>; <name>; <target_object>

#agentdb schema field-update [options] <object> <field>

Rename, retype or change a field. Every record is validated first; nothing changes on failure. Use --dry-run to preview.

Arguments: <object>; <field>

#agentdb schema field-remove [options] <object> <field>

Remove a field and its value from every record (needs --confirm; --dry-run previews)

Arguments: <object>; <field>

#agentdb schema unique-key-add [options] <object> <fields>

Declare fields that together identify a record (comma-separated, 2-5), usable as --match-on a,b,c

Arguments: <object>; <fields>

#agentdb schema unique-key-remove [options] <object> <fields>

Arguments: <object>; <fields>

#agentdb schema history [options]

Schema changes, newest first, with the definition before and after (find a change id to undo)

#agentdb schema undo [options] <change_id>

Reverse one schema change and restore archived values (--dry-run previews; --confirm when values are removed or can't all be restored)

Arguments: <change_id>

#agentdb records create [options] <object>

Arguments: <object>

#agentdb records list [options] <object>

List records of one object type (oldest first unless --sort); use query for filters

Arguments: <object>

#agentdb records delete [options] <id>

Delete one record (and its links and attachments; files are kept)

Arguments: <id>

#agentdb records delete-many [options] <object>

Delete up to 1000 records by --ids or by --where (needs --expected-count; try --dry-run first)

Arguments: <object>

#agentdb records count [options] <object>

Count records, optionally filtered and grouped (server-side; use aggregate for sums and averages)

Arguments: <object>

#agentdb records get [options] <id>

Arguments: <id>

#agentdb records update [options] <id>

Arguments: <id>

#agentdb records upsert [options] <object>

Bulk create-or-update from a JSONL file (one {"data":{...},"id"?,"version"?} per line) or a JSON array of such items. Sent in batches; each batch is one server transaction.

Arguments: <object>

#agentdb records upsert-one [options] <object>

Create-or-update one record by its match key, e.g. --match-on first_name,last_name,company --data '{"first_name":"Ann",...}'

Arguments: <object>

#agentdb records link-batch [options] <source_object> <relationship>

Bulk link from a JSONL file or JSON array; one {"source_record_id"|"source_key", "target_record_id"|"target_key", "client_ref"?} per line, where a key is {"<unique field>": value}. Sent in batches; each batch is one server transaction.

Arguments: <source_object>; <relationship>

#agentdb records link [options] <source_id> <relationship> <target_id>

Arguments: <source_id>; <relationship>; <target_id>

#agentdb records unlink [options] <link_id>

Arguments: <link_id>

#agentdb records related [options] <id> <relationship>

Arguments: <id>; <relationship>

#agentdb records archive [options] <ids...>

Archive records (left out of queries unless --archived / include_archived)

Arguments: <ids...>

#agentdb records unarchive [options] <ids...>

Bring archived records back into queries

Arguments: <ids...>

#agentdb records recycle-bin [options]

Deleted records that can still be restored

#agentdb records restore [options] <ids...>

Restore deleted records with their links and attachments

Arguments: <ids...>

#agentdb records purge [options] <ids...>

Delete records in the recycle bin for good (cannot be undone)

Arguments: <ids...>

#agentdb records duplicates [options]

Duplicate suggestions, e.g. --json '{"object":"contact","fields":["name"],"mode":"normalized"}'

#agentdb records merge-preview [options] <survivor_id> <merged_ids...>

What a merge would do: values, conflicts, links and files that move

Arguments: <survivor_id>; <merged_ids...>

#agentdb records merge [options]

Merge with explicit field_choices and versions from merge-preview

#agentdb records merge-undo [options] <merge_id>

Undo a merge while the merged records are in the recycle bin

Arguments: <merge_id>

#agentdb records merges [options]

Merge history

#agentdb files upload [options] <path>

Upload a file (streamed, checksum-verified) and optionally attach it to a record

Arguments: <path>

#agentdb files get [options] <file_id>

Arguments: <file_id>

#agentdb files list [options]

#agentdb files update [options] <file_id>

Rename a file or change its folder and tags (content is not changed)

Arguments: <file_id>

#agentdb files replace [options] <file_id> <path>

Upload new content for a file as a new version; attachments and file fields keep pointing at it

Arguments: <file_id>; <path>

#agentdb files versions [options] <file_id>

Version history of a file

Arguments: <file_id>

#agentdb files delete-version [options] <file_id> <version>

Delete an earlier version or cancel a pending replacement

Arguments: <file_id>; <version>

#agentdb files search [options] <query>

Find files by name or indexed text (exact substring; reports what was indexed)

Arguments: <query>

#agentdb files link [options] <file_id>

Signed, expiring link that opens the file without an API key (for dashboards and public apps)

Arguments: <file_id>

#agentdb files links [options] [file_id]

List signed file links (of one file, or of the workspace)

Arguments: [file_id]

#agentdb files unlink [options] <link_id>

Revoke a signed file link

Arguments: <link_id>

#agentdb files download [options] <file_id>

Arguments: <file_id>

#agentdb files attach [options] <file_id> <record_id>

Arguments: <file_id>; <record_id>

#agentdb files detach [options] <attachment_id>

Arguments: <attachment_id>

#agentdb files delete [options] <file_id>

Arguments: <file_id>

#agentdb aggregate [options]

Server-side count/sum/avg/min/max with optional group_by, e.g. --json '{"object":"invoice","group_by":[{"field":"issued_on","bucket":"month"}],"metrics":[{"op":"sum","field":"amount"}]}'

#agentdb query [options]

Run a typed query (see docs/QUICKSTART.md for the DSL)

#agentdb dashboards create [options]

Create a data link; prints data_url and a ready fetch loop (usage.js)

#agentdb dashboards list [options]

List dashboards (never shows links, keys or password hashes)

#agentdb dashboards update [options] <dashboard_id>

Change sharing settings; the link stays the same

Arguments: <dashboard_id>

#agentdb dashboards revoke [options] <dashboard_id>

Stop the link at once (also for viewers that are watching)

Arguments: <dashboard_id>

#agentdb dashboard-sites create [options]

Create a custom dashboard (nothing is live until deploy --publish)

#agentdb dashboard-sites list [options]

List custom dashboards with URL, mode and status

#agentdb dashboard-sites get [options] <site_id>

Versions, files, validation problems, sharing and recent events

Arguments: <site_id>

#agentdb dashboard-sites deploy [options] <site_id> <dir>

Upload a folder as a new version (index.html required; manifest.json for data sources); --publish makes it live

Arguments: <site_id>; <dir>

#agentdb dashboard-sites publish [options] <site_id>

Make a ready version live (an older version rolls back)

Arguments: <site_id>

#agentdb dashboard-sites share [options] <site_id>

Show sharing, or change it with --mode/--password-stdin/--expires-in-days

Arguments: <site_id>

#agentdb dashboard-sites invite [options] <site_id> <emails...>

Invite people by email (mode people); each gets a personal single-use link

Arguments: <site_id>; <emails...>

#agentdb dashboard-sites revoke-viewer [options] <site_id>

Remove one invited viewer

Arguments: <site_id>

#agentdb dashboard-sites unpublish [options] <site_id>

Take it offline; versions are kept

Arguments: <site_id>

#agentdb dashboard-sites revoke [options] <site_id>

Stop it for good, at once (also open pages)

Arguments: <site_id>

#agentdb dashboard-sites delete [options] <site_id>

Delete it and its stored files

Arguments: <site_id>

#agentdb dashboard-sites regenerate-link [options] <site_id>

Link mode: a new link (?k= key); the old link stops at once

Arguments: <site_id>

#agentdb dashboard-sites pause [options] <site_id>

Pause it: a neutral paused page, no data, every session ends

Arguments: <site_id>

#agentdb dashboard-sites resume [options] <site_id>

Resume a paused dashboard

Arguments: <site_id>

#agentdb access [options]

Who you are in the workspace (key, agent or member), your role and per object/field access

#agentdb docs [options]

API and schema docs generated from this workspace (what you may see)

#agentdb members list [options]

Owner, members and open invitations (owner/admin)

#agentdb members invite [options] <email>

Email a 7-day invitation link

Arguments: <email>

#agentdb members role [options] <member_id> <role>

Change a member's role

Arguments: <member_id>; <role>

#agentdb members remove [options] <member_id>

Remove a member or withdraw an invitation

Arguments: <member_id>

#agentdb agents list [options]

Agents with role, status and who granted them

#agentdb agents create [options] <name>

Create an agent and print its key once

Arguments: <name>

#agentdb agents key [options] <agent_id>

Issue another key for an agent (printed once)

Arguments: <agent_id>

#agentdb agents revoke [options] <agent_id>

Revoke an agent, its keys and the agents it delegated to

Arguments: <agent_id>

#agentdb permissions list [options]

All rules (owner/admin)

#agentdb permissions set [options] <subject> <object> [field]

Create or replace a rule; subject is role:viewer, role:editor, user:<acc_…> or agent:<agt_…>

Arguments: <subject>; <object>; [field]

#agentdb permissions remove [options] <rule_id>

Remove a rule

Arguments: <rule_id>

#agentdb webhooks create [options] <url>

Register an https endpoint; prints the signing secret ONCE

Arguments: <url>

#agentdb webhooks list [options]

List webhooks (never shows secrets)

#agentdb webhooks get [options] <webhook_id>

One webhook with delivery counts

Arguments: <webhook_id>

#agentdb webhooks update [options] <webhook_id>

Change url, events, objects or name

Arguments: <webhook_id>

#agentdb webhooks pause [options] <webhook_id>

Stop sending (events keep queueing)

Arguments: <webhook_id>

#agentdb webhooks resume [options] <webhook_id>

Send again (also after an automatic disable)

Arguments: <webhook_id>

#agentdb webhooks delete [options] <webhook_id>

Delete; queued deliveries are cancelled

Arguments: <webhook_id>

#agentdb webhooks rotate-secret [options] <webhook_id>

New secret (printed once); the old one also signs during the grace period

Arguments: <webhook_id>

#agentdb webhooks test [options] <webhook_id>

Queue a signed webhook.test delivery

Arguments: <webhook_id>

#agentdb webhooks deliveries [options]

Delivery log, newest first (status: pending, delivering, succeeded, dead, cancelled)

#agentdb webhooks redeliver [options] <delivery_id>

Send a finished delivery again (same event id)

Arguments: <delivery_id>

#agentdb jobs create [options]

e.g. --json '{"name":"Daily digest","schedule":{"type":"cron","expression":"0 9 * * 1-5","timezone":"Europe/Amsterdam"},"action":{"type":"create_record","object":"agent_task","data":{"title":"Write digest"}}}'

#agentdb jobs list [options]

#agentdb jobs get [options] <job_id>

One job with its next 5 run times

Arguments: <job_id>

#agentdb jobs update [options] <job_id>

Change name, schedule, action, missed or max_attempts

Arguments: <job_id>

#agentdb jobs pause [options] <job_id>

Run nothing until resumed

Arguments: <job_id>

#agentdb jobs resume [options] <job_id>

Resume from now (missed times are not made up)

Arguments: <job_id>

#agentdb jobs cancel [options] <job_id>

Cancel for good

Arguments: <job_id>

#agentdb jobs run [options] <job_id>

Run once now

Arguments: <job_id>

#agentdb jobs runs [options]

Run history, newest first

#agentdb jobs retry [options] <run_id>

Retry a dead run

Arguments: <run_id>

#agentdb jobs preview [options]

Next times of a schedule without creating it (also checks a recurrence field value)

#agentdb flows create [options]

Create a flow (try it first with flows test)

#agentdb flows list [options]

#agentdb flows get [options] <flow_id>

One flow with versions and run counts

Arguments: <flow_id>

#agentdb flows update [options] <flow_id>

Change a flow (keeps a new version)

Arguments: <flow_id>

#agentdb flows pause [options] <flow_id>

Arguments: <flow_id>

#agentdb flows resume [options] <flow_id>

Arguments: <flow_id>

#agentdb flows delete [options] <flow_id>

Arguments: <flow_id>

#agentdb flows test [options]

Dry run, e.g. --json '{"flow_id":"flw_…","record_id":"rec_…"}' (nothing is written or sent)

#agentdb flows runs [options]

Flow run history (audit)

#agentdb channels create [options] <name>

Arguments: <name>

#agentdb channels list [options]

#agentdb channels get [options] <channel_id>

Arguments: <channel_id>

#agentdb channels update [options] <channel_id>

e.g. --json '{"topic":"…","retention_days":30,"archived":false}'

Arguments: <channel_id>

#agentdb channels add-members [options] <channel_id> <identities>

Arguments: <channel_id>; <identities>

#agentdb channels members [options] <channel_id>

Members with read receipts and ack offsets

Arguments: <channel_id>

#agentdb channels remove-member [options] <channel_id> <identity>

Remove a member (your own identity = leave)

Arguments: <channel_id>; <identity>

#agentdb channels subscribe [options] <channel_id>

Join a public channel or change your notify level

Arguments: <channel_id>

#agentdb channels direct [options] <identity>

Open the direct channel with an agent or human

Arguments: <identity>

#agentdb channels post [options] <channel_id> [body]

Arguments: <channel_id>; [body]

#agentdb channels messages [options] <channel_id>

Arguments: <channel_id>

#agentdb channels pending [options] <channel_id>

Messages after your ack offset (they come back until acked)

Arguments: <channel_id>

#agentdb channels ack [options] <channel_id> <seq>

Arguments: <channel_id>; <seq>

#agentdb channels purge [options] <channel_id>

Apply the retention now (moderators)

Arguments: <channel_id>

#agentdb channels message [options] <message_id>

One message (with history for the author and moderators)

Arguments: <message_id>

#agentdb channels edit [options] <message_id> <body>

Arguments: <message_id>; <body>

#agentdb channels delete-message [options] <message_id>

Arguments: <message_id>

#agentdb channels search [options] <query>

Arguments: <query>

#agentdb inbox list [options]

#agentdb inbox read [options] <item_id>

Mark an item read

Arguments: <item_id>

#agentdb inbox unread [options] <item_id>

Mark an item unread

Arguments: <item_id>

#agentdb inbox done [options] <item_id>

Mark an item done

Arguments: <item_id>

#agentdb inbox notify [options] <identity> <title>

Arguments: <identity>; <title>

#agentdb conversations create [options] <subject>

Arguments: <subject>

#agentdb conversations list [options]

#agentdb conversations get [options] <conversation_id>

The conversation with every entry

Arguments: <conversation_id>

#agentdb conversations reply [options] <conversation_id> <body>

Arguments: <conversation_id>; <body>

#agentdb conversations assign [options] <conversation_id> <identity>

Hand over to an agent or human ('none' = back to the queue)

Arguments: <conversation_id>; <identity>

#agentdb conversations claim [options] <conversation_id>

Arguments: <conversation_id>

#agentdb conversations resolve [options] <conversation_id>

Arguments: <conversation_id>

#agentdb approvals submit [options]

Propose a write, e.g. --json '{"operation":"records.update","params":{"record":"rec_…"},"input":{"version":3,"data":{"amount":5000}},"justification":"…"}'

#agentdb approvals list [options]

Proposals, newest first

#agentdb approvals get [options] <approval_id>

One proposal: input, preview, decisions, result

Arguments: <approval_id>

#agentdb approvals approve [options] <approval_id>

Arguments: <approval_id>

#agentdb approvals reject [options] <approval_id>

Arguments: <approval_id>

#agentdb approvals request-changes [options] <approval_id>

Arguments: <approval_id>

#agentdb approvals cancel [options] <approval_id>

Withdraw a pending proposal

Arguments: <approval_id>

#agentdb approvals policies [options] [command]

Which writes need approval and who decides (owners and admins)

#agentdb approvals policies list [options]

#agentdb approvals policies create [options]

e.g. --json '{"name":"Large amounts","operations":["records.update"],"objects":["invoice"],"condition":{"field":"amount","op":"gt","value":1000}}'

#agentdb approvals policies update [options] <policy_id>

Change a policy (include its current version)

Arguments: <policy_id>

#agentdb approvals policies delete [options] <policy_id>

Arguments: <policy_id>

#agentdb calendar create [options]

Create a calendar: {name, timezone, owner?, writers?}

#agentdb calendar list [options]

#agentdb calendar get [options] <calendar_id>

Arguments: <calendar_id>

#agentdb calendar update [options] <calendar_id>

Arguments: <calendar_id>

#agentdb calendar delete [options] <calendar_id>

Delete a calendar with its events

Arguments: <calendar_id>

#agentdb calendar event-create [options]

Book an event (see docs/CALENDAR.md for start/end formats, recurrence, participants, reminders)

#agentdb calendar event-get [options] <event_id>

Arguments: <event_id>

#agentdb calendar event-update [options] <event_id>

Change an event; add "occurrence" to change one occurrence

Arguments: <event_id>

#agentdb calendar event-delete [options] <event_id>

Arguments: <event_id>

#agentdb calendar respond [options] <event_id> <response>

accepted, declined, tentative or needs_action (for yourself)

Arguments: <event_id>; <response>

#agentdb calendar events [options]

Occurrences between --from and --to (RFC 3339)

#agentdb calendar availability [options]

Free-busy for calendars and participants

#agentdb calendar slots [options]

Find free meeting slots

#agentdb calendar expand [options]

Expand an RRULE into occurrences

#agentdb calendar export [options] <calendar_id>

iCalendar (.ics) export (JSON with an ics field)

Arguments: <calendar_id>

#agentdb calendar import [options] <calendar_id> <ics_file>

Import an .ics file (max 1 MiB)

Arguments: <calendar_id>; <ics_file>

#agentdb knowledge create [options]

Store an item with sources

#agentdb knowledge get [options] <item_id>

Arguments: <item_id>

#agentdb knowledge list [options]

#agentdb knowledge search [options] <query>

Arguments: <query>

#agentdb knowledge update [options] <item_id>

New version (a proposal on trusted items unless you are an owner/admin)

Arguments: <item_id>

#agentdb knowledge propose [options] <item_id>

Propose a change or contradiction with a reason

Arguments: <item_id>

#agentdb knowledge review [options] <item_id> <version> <decision>

Owners/admins: accept or reject a proposal

Arguments: <item_id>; <version>; <decision>

#agentdb knowledge versions [options] <item_id>

Arguments: <item_id>

#agentdb knowledge provenance [options] <item_id>

Arguments: <item_id>

#agentdb knowledge delete [options] <item_id>

Archive (default); --purge --confirm removes it for good

Arguments: <item_id>

#agentdb connectors types [options]

Connector types on this server

#agentdb connectors create [options]

Configure a connector (credentials are shown never again)

#agentdb connectors list [options]

#agentdb connectors get [options] <connector_id>

Arguments: <connector_id>

#agentdb connectors update [options] <connector_id>

Change config, rotate credentials, pause/resume

Arguments: <connector_id>

#agentdb connectors delete [options] <connector_id>

Arguments: <connector_id>

#agentdb connectors sync [options] <connector_id>

Run a sync now

Arguments: <connector_id>

#agentdb connectors rotate-inbound-secret [options] <connector_id>

Arguments: <connector_id>

#agentdb connectors send [options] <connector_id>

Perform an allowed action, e.g. --json '{"action":"notify","payload":{...}}'

Arguments: <connector_id>

#agentdb connectors runs [options]

Sync runs, newest first

#agentdb connectors inbound-events [options]

#agentdb import [options] <file>

Import a CSV/JSON/JSONL file on the server. Without --object/--match-on (or with --inspect) it only inspects: columns, types, match keys, spelling variants and suggested normalize rules. Run with --dry-run first; --normalize applies only the rules you give and reports every changed cell.

Arguments: <file>

#agentdb capabilities [options]

What the server offers (versions, MCP tools and when each was added, limits). No login needed. Find an operation: --search "attach a file", --category files, --operation files_attach, --index (all categories).

#agentdb search query [options] <object> <words...>

Records containing every word (--any: at least one); ranked, with highlights

Arguments: <object>; <words...>

#agentdb search configure [options] <object>

Enable (and build) the full-text index of an object; --disable removes it

Arguments: <object>

#agentdb search config [options]

Objects with a full-text index

#agentdb search semantic [options] <object> <words...>

Semantic ranking (FEATURE_DISABLED unless the operator enabled a provider)

Arguments: <object>; <words...>

#agentdb lifecycle info [options] [object]

Arguments: [object]

#agentdb lifecycle retention [options] <object> <days>

Days deleted records stay restorable (0 = permanent deletes, 'default' = 30)

Arguments: <object>; <days>

#agentdb integrity [options]

Dry-run integrity report: required relationships and records missing their link

#agentdb relationships require [options] <object> <relationship>

Make a relationship required (--off: optional again)

Arguments: <object>; <relationship>

#agentdb views create [options]

e.g. --json '{"name":"open_deals","object":"deal","visibility":"workspace","definition":{"where":{"all":[…]}}}'

#agentdb views list [options]

#agentdb views get [options] <view>

A view with its versions

Arguments: <view>

#agentdb views run [options] <view>

Arguments: <view>

#agentdb views update [options] <view>

Needs the current version, e.g. --json '{"version":1,"name":"x"}'

Arguments: <view>

#agentdb views delete [options] <view>

Arguments: <view>

#agentdb packages install [options] [manifest_file]

Install or upgrade a package from a JSON manifest, or --template crm|agent_crm|agent_todos|agent_kanban|agent_projects|support|recruit|property|inventory|assets

Arguments: [manifest_file]

#agentdb packages list [options]

Installed packages and built-in templates

#agentdb packages get [options] <name>

A package's or template's manifest

Arguments: <name>

#agentdb packages uninstall [options] <name>

Forget a package (and its views, flows and approval policies); objects, fields and records stay

Arguments: <name>

#agentdb cms configure [options]

Make an object a content type, e.g. --json '{"object":"cms_post","public_fields":["title","slug","body"]}'

#agentdb cms types [options]

Content types and their delivery settings

#agentdb cms unconfigure [options] <object>

Stop delivering an object (records and revisions stay)

Arguments: <object>

#agentdb cms publish [options] <record>

Publish (or re-publish) an item that is in review

Arguments: <record>

#agentdb cms unpublish [options] <record>

Arguments: <record>

#agentdb cms schedule [options] <record> <at>

Publish at an RFC 3339 time (a scheduler job)

Arguments: <record>; <at>

#agentdb cms revisions [options] <record>

Arguments: <record>

#agentdb cms revision [options] <record> <version>

Arguments: <record>; <version>

#agentdb cms diff [options] <record> <from> [to]

Arguments: <record>; <from>; [to]

#agentdb cms restore [options] <record> <revision>

Arguments: <record>; <revision>

#agentdb cms token-create [options] <label>

A read-only delivery token for a website (shown once)

Arguments: <label>

#agentdb cms tokens [options]

#agentdb cms token-revoke [options] <id>

Arguments: <id>

#agentdb forms create [options]

Create a form from a JSON definition (the token is shown once)

#agentdb forms list [options]

#agentdb forms get [options] <form>

Arguments: <form>

#agentdb forms update [options] <form>

Needs the current version, e.g. --json '{"version":1,"status":"closed"}'

Arguments: <form>

#agentdb forms rotate-token [options] <form>

Arguments: <form>

#agentdb forms delete [options] <form>

Arguments: <form>

#agentdb forms submissions [options]

#agentdb forms release [options] <submission>

Arguments: <submission>

#agentdb forms discard [options] <submission>

Arguments: <submission>

#agentdb content dispatch [options] <record>

Hand a post to the operator's outbound publisher (off by default)

Arguments: <record>

#agentdb indexes create [options] <name> <object> <field>

Arguments: <name>; <object>; <field>

#agentdb indexes list [options]

#agentdb indexes drop [options] <name>

Arguments: <name>

#agentdb data-export [options]

Download your data: records, links, schema and file metadata (ndjson), or --bundle (.tar.gz with file bytes). Owners and admins.

#agentdb login [options]

Log in through the browser (or --api-key) and save the key in the OS credential store

#agentdb logout [options]

Remove the saved key; a key issued by agentdb login is also revoked on the server

#agentdb whoami [options]

Show the workspace and key the CLI uses (checks it with the server)

#agentdb status [options]

Login, server and MCP setup at a glance (never prints secrets)

#agentdb mcp install [options] <client>

Register AgentDB in an AI tool (claude | codex | cursor | vscode | claude-desktop | windsurf | gemini); the tool runs agentdb mcp proxy, so no key is written to its config

Arguments: <client>

#agentdb mcp remove [options] <client>

Remove AgentDB from an AI tool (other servers and settings are kept)

Arguments: <client>

#agentdb mcp list [options]

Show which AI tools have AgentDB configured

#agentdb mcp proxy [options]

stdio MCP server forwarding to AgentDB with the saved login (started by the AI tool, not by you)

#agentdb skill print [options]

Print SKILL.md

#agentdb skill install [options] <client>

Install SKILL.md for Claude Code (~/.claude/skills/agentdb/SKILL.md)

Arguments: <client>