Runtime settings catalog

Inspect catalex settings --json for your release’s effective values, validated bounds and editability. This reference records the published setting names and defaults checked on October 24, 2026; supported fields vary by version. It deliberately contains no machine-specific values or secrets.

Use capacity and storage for examples. Set an editable key with catalex settings set KEY=JSON_VALUE; null restores its default. Restart services means running services are restarted to apply the change; controller means a controller restart is required; none applies without a restart. Read-only network fields use setup or the endpoint workflow, not a raw settings write.

The live catalog supplies ranges because related limits must remain compatible: for example, a ping interval must be shorter than its receiving connection’s idle timeout. A maximum allowed by validation is not a recommended workload. Increase capacity only after checking memory, disk and provider limits.

Workload and history

Bound simultaneous work, waiting inputs and history size. Raise capacity only when the computer can support it.

Key	Purpose	Default	Unit	Change / restart
limits.max_active_turns	How many agent turns may run at the same time	4	count	Editable / services
limits.max_warm_runtimes	How many idle agent runtimes stay ready for quick replies	8	count	Editable / services
limits.max_queued_inputs_per_session	How many messages may wait for one chat before new ones are refused	20	count	Editable / services
limits.bridge_spool_mib	Disk space for messages held while the messaging connection is down	256	MiB	Editable / services
limits.local_backlog_mib	Disk space for messages waiting to be delivered to agents	1024	MiB	Editable / services
limits.bootstrap_messages	How many earlier messages a new agent session is given as context	100	count	Editable / services
limits.inventory_probe_timeout_ms	How long to wait when checking an agent's installed version and models	10000	ms	Editable / services
limits.inventory_max_models	How many models to list per agent	200	count	Editable / services

Memory

Bound output buffering and in-memory records when agents or histories become large.

Key	Purpose	Default	Unit	Change / restart
limits.memory.output_buffer_bytes	Memory kept for each running agent's output	131072	bytes	Editable / services
limits.memory.peer_tracked_commands	How many peer commands are remembered while awaiting a result	128	count	Editable / services
limits.memory.peer_received_frames	How many received peer messages may be held in memory	100	count	Editable / services
limits.memory.peer_received_bytes	Memory allowed for received peer messages	2097152	bytes	Editable / services

Recovery

Bound attempts and delay after interrupted or failed sessions; preserve a finite recovery budget.

Key	Purpose	Default	Unit	Change / restart
limits.recovery.max_attempts	How many times a failed turn is retried before giving up	3	count	Editable / services
limits.recovery.series_reset_seconds	How long without failures before the retry count starts over	1800	s	Editable / services
limits.recovery.scan_page_size	How many interrupted turns are checked at once on startup	32	count	Editable / services
limits.recovery.max_same_session_retries	How many retries reuse the same agent session before starting fresh	2	count	Editable / services
limits.recovery.backoff_seconds	Seconds to wait before each successive retry	[5,30]	s	Editable / services
limits.recovery.prompt_resends	How many times a message is resent to an agent that did not answer	2	count	Editable / services

Messaging connection

Tune bounded connection, replay and acknowledgement behavior for a measured network problem.

Key	Purpose	Default	Unit	Change / restart
limits.peer.native_session_sync_interval_ms	How often chats started in an agent's own app are synced	60000	ms	Editable / services
limits.peer.native_session_probe_timeout_ms	How long a check for an agent's own chats may take	30000	ms	Editable / services
limits.peer.native_session_page_size	How many of an agent's own chats are read per request	100	count	Editable / services
limits.peer.native_session_max_pages	How many pages of an agent's own chats are read per sync	1000	count	Editable / services
limits.peer.call_timeout_ms	How long a peer call may take before it fails	5000	ms	Editable / services
limits.peer.command_timeout_ms	How long a peer command may take before it fails	10000	ms	Editable / services
limits.peer.handshake_timeout_ms	How long a new peer connection may take to be accepted	5000	ms	Editable / services
limits.peer.write_timeout_ms	How long sending one message to a peer may take	5000	ms	Editable / services
limits.peer.read_idle_timeout_ms	How long a peer connection may stay silent before it is dropped	30000	ms	Editable / services
limits.peer.ping_interval_ms	How often a peer connection is checked to be alive	10000	ms	Editable / services
limits.peer.poll_interval_ms	Fallback for new work; a store commit wakes chat links at once	250	ms	Editable / services
limits.peer.reconnect_initial_ms	How long to wait before the first reconnect to a peer	1000	ms	Editable / services
limits.peer.reconnect_max_ms	The longest wait between reconnect attempts to a peer	30000	ms	Editable / services
limits.peer.reconnect_reset_ms	How long a connection must hold before reconnect waits reset	30000	ms	Editable / services
limits.peer.shutdown_timeout_ms	How long peer connections get to close cleanly on shutdown	5000	ms	Editable / services
limits.peer.claim_lease_ms	How long a claimed message is reserved before others may take it	15000	ms	Editable / services
limits.peer.claim_batch	How many messages are claimed from a peer at once	32	count	Editable / services
limits.peer.session_page_size	How many chats are listed per peer request	32	count	Editable / services
limits.peer.cursor_ttl_ms	How long a peer paging cursor stays valid	600000	ms	Editable / services
limits.peer.received_page_bytes	The largest page of data accepted from a peer in one reply	524288	bytes	Editable / services
limits.peer.control_queue_bytes	Memory for control messages queued to a peer	1048576	bytes	Editable / services
limits.peer.data_queue_bytes	Memory for chat data queued to a peer	2097152	bytes	Editable / services
limits.peer.queue_frames	How many messages may queue for one peer connection	32	count	Editable / services
limits.peer.accepted_links	How many peer connections are accepted at once	16	count	Editable / services
limits.peer.follow_batch	How many updates are followed from a peer per round	32	count	Editable / services
limits.peer.target_fetch_timeout_ms	How long fetching a message's target may take	10000	ms	Editable / services
limits.peer.target_fetch_batch	How many targets are fetched per request	20	count	Editable / services
limits.peer.target_fetches	How many target fetches may run at the same time	8	count	Editable / services

Agent communication

Bound communication queues, output, retries and bridge liveness; avoid using longer timeouts to hide an offline peer.

Key	Purpose	Default	Unit	Change / restart
limits.bridge.handshake_timeout_ms	How long the messaging connection may take to connect	5000	ms	Editable / services
limits.bridge.write_timeout_ms	How long sending one message over the messaging connection may take	5000	ms	Editable / services
limits.bridge.read_idle_timeout_ms	How long the messaging connection may stay silent before it is dropped	30000	ms	Editable / services
limits.bridge.ping_interval_ms	How often the messaging connection is checked to be alive	10000	ms	Editable / services
limits.bridge.poll_interval_ms	Fallback for forwarding work; a journal commit forwards at once	100	ms	Editable / services
limits.bridge.control_page_size	How many control messages are read per request	32	count	Editable / services
limits.bridge.reconnect_initial_ms	How long to wait before the first hosted reconnect	1000	ms	Editable / services
limits.bridge.reconnect_max_ms	The longest wait between hosted reconnect attempts	30000	ms	Editable / services
limits.bridge.shutdown_timeout_ms	How long the messaging connection gets to close cleanly on shutdown	10000	ms	Editable / services
limits.bridge.history_timeout_ms	How long loading chat history for the hosted side may take	15000	ms	Editable / services
limits.bridge.history_requests	How many history requests may run at the same time	32	count	Editable / services
limits.bridge.native_open_timeout_ms	How long opening an agent's own chat may take	60000	ms	Editable / services
limits.bridge.cancel_timeout_ms	How long cancelling a running turn may take	250	ms	Editable / services
limits.bridge.process_exit_wait_ms	How long to wait for an agent process to exit	5000	ms	Editable / services
limits.bridge.process_poll_interval_ms	How often a running agent process is checked	10	ms	Editable / services
limits.bridge.tool_call_timeout_ms	How long an agent's tool call may take before it fails	20000	ms	Editable / services
limits.bridge.io_buffer_bytes	Memory used to read and write agent output	4096	bytes	Editable / services
limits.bridge.journal_total_mib	Total disk space for the delivery journal	256	MiB	Editable / services
limits.bridge.journal_tenant_mib	Disk space the delivery journal may use per account	100	MiB	Editable / services
limits.bridge.tool_refusals	How many refused tool calls are tolerated before a turn stops	4	count	Editable / services
limits.bridge.routine_messages_per_second	How many routine messages an agent may send each second	4	count	Editable / services
limits.bridge.tool_connections	How many tool connections may be open at once	16	count	Editable / services
limits.bridge.journal_lock_timeout_ms	How long to wait for the delivery journal lock	50	ms	Editable / services
limits.bridge.journal_disk_reserve_mib	Free disk space kept back from the delivery journal	8	MiB	Editable / services

Conversation gateway

Bound incoming sessions, delivery, decisions and attachment requests.

Key	Purpose	Default	Unit	Change / restart
limits.gateway.read_header_timeout_ms	How long a client may take to send request headers	5000	ms	Editable / services
limits.gateway.idle_timeout_ms	How long an idle client connection is kept open	30000	ms	Editable / services
limits.gateway.shutdown_timeout_ms	How long the gateway gets to stop cleanly	10000	ms	Editable / services
limits.gateway.drain_timeout_ms	How long in-flight work may finish before the gateway stops	15000	ms	Editable / services
limits.gateway.remote_history_timeout_ms	How long loading remote chat history may take	15000	ms	Editable / services
limits.gateway.cursor_ttl_ms	How long a paging cursor stays valid	300000	ms	Editable / services
limits.gateway.archive_timeout_ms	How long archiving a chat may take	10000	ms	Editable / services
limits.gateway.retention_interval_ms	How often old data is checked for cleanup	2000	ms	Editable / services
limits.gateway.retention_high_percent	Storage fullness that starts removing old data	70	percent	Editable / services
limits.gateway.retention_low_percent	Storage fullness at which cleanup of old data stops	60	percent	Editable / services
limits.gateway.dispatch_interval_ms	How often waiting messages are handed to agents	250	ms	Editable / services
limits.gateway.dispatch_pass_timeout_ms	How long one delivery pass may take	5000	ms	Editable / services
limits.gateway.drain_poll_interval_ms	How often shutdown checks whether work has finished	100	ms	Editable / services
limits.gateway.bridge_handshake_timeout_ms	How long a bridge connection may take to be accepted	5000	ms	Editable / services
limits.gateway.bridge_write_timeout_ms	How long sending one message to a bridge may take	5000	ms	Editable / services
limits.gateway.bridge_ping_interval_ms	How often a bridge connection is checked to be alive	10000	ms	Editable / services
limits.gateway.bridge_read_idle_timeout_ms	How long a bridge may stay silent before it is dropped	30000	ms	Editable / services
limits.gateway.bridge_final_wait_ms	How long a success reaction waits for its turn's end	5000	ms	Editable / services
limits.gateway.bridge_queue_frames	How many messages may queue for one bridge connection	32	count	Editable / services
limits.gateway.bridge_pending_finals	How many success reactions may wait for their turn's end	32	count	Editable / services
limits.gateway.bridge_io_buffer_bytes	Memory used to read and write bridge traffic	4096	bytes	Editable / services
limits.gateway.event_cursors	How many live event streams may be open at once	256	count	Editable / services
limits.gateway.session_cursors	How many chat list cursors may be open at once	1024	count	Editable / services
limits.gateway.event_page_size	How many events are returned per request	10	count	Editable / services
limits.gateway.catalogue_page_size	How many catalogue entries are returned per request	100	count	Editable / services
limits.gateway.history_page_size	How many history messages are returned per request	20	count	Editable / services
limits.gateway.remote_history_skip_pages	How many remote history pages may be skipped when catching up	5	count	Editable / services
limits.gateway.idle_tracked_sessions	How many chats are tracked for idle detection	4096	count	Editable / services
limits.gateway.analytics_sample_records	How many records are sampled for analytics	2000	count	Editable / services
limits.gateway.analytics_transcript_messages	How many messages per chat are read for analytics	100	count	Editable / services

Chat client connections

Bound chat requests, history pages and client handshakes so a slow client cannot hold resources indefinitely.

Key	Purpose	Default	Unit	Change / restart
limits.console.gateway_timeout_ms	How long the dashboard waits for the gateway to answer	10000	ms	Editable / services
limits.console.links	How many dashboard connections may be open at once	16	count	Editable / services
limits.console.link_handshake_timeout_ms	How long a dashboard connection may take to be accepted	10000	ms	Editable / services
limits.console.link_write_timeout_ms	How long sending one message to the dashboard may take	5000	ms	Editable / services
limits.console.link_ping_interval_ms	How often a dashboard connection is checked to be alive	10000	ms	Editable / services
limits.console.link_idle_timeout_ms	How long a dashboard connection may stay silent before it is dropped	30000	ms	Editable / services
limits.console.answer_replay_timeout_ms	How long a given answer is kept for replay after a reconnect	15000	ms	Editable / services
limits.console.config_apply_timeout_ms	How long applying a settings change may take	2000	ms	Editable / services
limits.console.config_apply_interval_ms	How often a pending settings change is retried	200	ms	Editable / services
limits.console.remembered_scopes	How many permission choices are remembered	10000	count	Editable / services
limits.console.page_size	How many items the dashboard loads per page	50	count	Editable / services
limits.console.attachment_status_batch	How many attachment statuses are checked per request	50	count	Editable / services

Management operations

Bound setup, discovery, service checks and management requests on the agent computer.

Key	Purpose	Default	Unit	Change / restart
limits.controller.endpoint_operations	How many endpoint operations may run at the same time	128	count	Editable / services
limits.controller.start_timeout_ms	How long starting CatalEx may take	300000	ms	Editable / services
limits.controller.cleanup_timeout_ms	How long cleaning up leftovers may take	20000	ms	Editable / services
limits.controller.gateway_ready_timeout_ms	How long to wait for the gateway to become ready	90000	ms	Editable / services
limits.controller.gateway_probe_timeout_ms	How long one gateway readiness check may take	3000	ms	Editable / services
limits.controller.gateway_probe_interval_ms	How often the gateway is checked for readiness	1000	ms	Editable / services
limits.controller.shutdown_timeout_ms	How long CatalEx gets to stop cleanly	15000	ms	Editable / services
limits.controller.drain_timeout_ms	How long the gateway gets to finish work before stopping	20000	ms	Editable / services
limits.controller.control_timeout_ms	How long a request to the controller may take	5000	ms	Editable / services
limits.controller.down_timeout_ms	How long stopping CatalEx may take	25000	ms	Editable / services
limits.controller.down_poll_interval_ms	How often shutdown checks that services have stopped	250	ms	Editable / services
limits.controller.control_read_header_timeout_ms	How long a control client may take to send request headers	3000	ms	Editable / services
limits.controller.console_read_header_timeout_ms	How long a browser may take to send request headers	5000	ms	Editable / services
limits.controller.console_idle_timeout_ms	How long an idle browser connection is kept open	60000	ms	Editable / services
limits.controller.media_reconcile_interval_ms	How often stored media is checked against chats	3600000	ms	Editable / services
limits.controller.socket_probe_timeout_ms	How long checking a local socket may take	200	ms	Editable / services
limits.controller.orphan_stop_timeout_ms	How long to wait when stopping leftover processes	120000	ms	Editable / services
limits.controller.stats_interval_ms	How often resource usage is sampled	30000	ms	Editable / services
limits.controller.stats_docker_timeout_ms	How long reading container usage may take	5000	ms	Editable / services
limits.controller.stats_chats	How many chats are included in usage statistics	500	count	Editable / services
limits.controller.stats_processes	How many processes are included in usage statistics	256	count	Editable / services
limits.controller.slack_setup_timeout_ms	How long one Slack setup step on the messaging service may take	25000	ms	Editable / services

Conversation storage

Bound database retries, maintenance and paging. Change only for a diagnosed storage problem.

Key	Purpose	Default	Unit	Change / restart
limits.store.query_timeout_ms	How long a database query may take	5000	ms	Editable / services
limits.store.annotation_timeout_ms	How long saving a message annotation may take	2000	ms	Editable / services
limits.store.control_expiry_ms	How long an unanswered control request stays valid	60000	ms	Editable / services
limits.store.sqlite_busy_timeout_ms	How long to wait for a busy local database	5000	ms	Editable / services
limits.store.postgres_connect_timeout_ms	How long connecting to the PostgreSQL database may take	5000	ms	Editable / services
limits.store.max_open_conns	The most database connections open at once	10	count	Editable / services
limits.store.max_idle_conns	The most idle database connections kept ready	2	count	Editable / services
limits.store.touch_buffer	How many chat activity updates are batched before saving	1024	count	Editable / services
limits.store.evict_batch	How many old records are removed per cleanup step	500	count	Editable / services
limits.store.settlement_reserve_percent	Share of storage reserved for finishing in-flight messages	5	percent	Editable / services
limits.store.settled_control_retention_ms	How long completed control requests are kept	604800000	ms	Editable / services
limits.store.handover_batch	How many records move to a new owner per step	256	count	Editable / services
limits.store.dispatch_page_size	How many waiting messages are read per delivery pass	32	count	Editable / services
limits.store.command_page_size	How many commands are read per request	32	count	Editable / services

Native provider sessions

Bound startup, decisions, native output and cancellation for the installed AI tools.

Key	Purpose	Default	Unit	Change / restart
limits.native.version_probe_timeout_ms	How long checking an agent's version may take	5000	ms	Editable / services
limits.native.login_probe_timeout_ms	How long checking an agent's sign-in may take	10000	ms	Editable / services
limits.native.reap_timeout_ms	How long to wait for a finished agent process to be cleaned up	1000	ms	Editable / services
limits.native.exit_grace_ms	How long an agent gets to exit after being asked to stop	500	ms	Editable / services
limits.native.stop_grace_ms	How long an agent gets to stop before it is forced	1000	ms	Editable / services
limits.native.kill_wait_ms	How long to wait after forcing an agent process to end	1000	ms	Editable / services
limits.native.cancel_timeout_ms	How long cancelling an agent turn may take	250	ms	Editable / services
limits.native.write_timeout_ms	How long sending input to an agent may take	1000	ms	Editable / services
limits.native.completion_wait_ms	How long to wait for an agent to report a turn finished	2000	ms	Editable / services
limits.native.call_match_timeout_ms	How long to match an agent's tool call to its result	1000	ms	Editable / services
limits.native.permission_match_timeout_ms	How long to match a permission request to an agent's tool call	2000	ms	Editable / services
limits.native.turn_process_stop_timeout_ms	How long stopping an agent process after a turn may take	2000	ms	Editable / services
limits.native.decision_expiry_ms	How long a permission question waits for an answer	1800000	ms	Editable / services
limits.native.group_poll_interval_ms	How often an agent's process group is checked	10	ms	Editable / services
limits.native.probe_pipe_wait_ms	How long to wait for an agent's probe output	100	ms	Editable / services
limits.native.permission_hook_io_timeout_ms	How long the permission check may take to read or write	5000	ms	Editable / services
limits.native.permission_hook_margin_ms	Extra time allowed on top of the permission answer wait	60000	ms	Editable / services

Audio

Bound audio bytes and duration before transcription or other processing.

Key	Purpose	Default	Unit	Change / restart
limits.audio.max_bytes	The largest voice message accepted	10485760	bytes	Editable / services
limits.audio.max_duration_ms	The longest voice message accepted	300000	ms	Editable / services

Attachments

Balance attachment size, count, retention and total disk use. These settings do not erase downloaded recipient copies.

Key	Purpose	Default	Unit	Change / restart
limits.artifacts.retention_ms	How long generated files are kept	604800000	ms	Editable / services
limits.artifacts.max_spool_bytes	Disk space for generated files awaiting delivery	268435456	bytes	Editable / services
limits.artifacts.max_attachments	How many attachments one message may carry	4	count	Editable / services
limits.artifacts.max_file_bytes	The largest single attachment accepted	10485760	bytes	Editable / services
limits.artifacts.upload_timeout_ms	How long uploading one agent file to the messaging service may take	120000	ms	Editable / services

Agent messaging

Bound cross-agent message work and delivery records. These limits do not enable a destination; the sender’s allowlist does.

Key	Purpose	Default	Unit	Change / restart
limits.delegation.ref_cache_entries	How many routes to other relays' contacts are remembered	10000	count	Editable / services
limits.delegation.resolve_timeout_ms	How long to wait for the messaging service to route a delegation	5000	ms	Editable / services
limits.delegation.outbox_poll_interval_ms	How often delegation messages for the messaging service are collected	250	ms	Editable / services
limits.delegation.outbox_poll_wait_ms	How long a collection of delegation messages waits for one to send	2000	ms	Editable / services
limits.delegation.link_report_timeout_ms	How long the messaging service link counts as up without word from it	3000	ms	Editable / services

Controller process

Tune the management process’s own request and operation budgets. These require a controller restart rather than only a service restart.

Key	Purpose	Default	Unit	Change / restart
controller.limits.controller.endpoint_operations	How many endpoint operations may run at the same time	128	count	Editable / controller
controller.limits.controller.start_timeout_ms	How long starting CatalEx may take	300000	ms	Editable / controller
controller.limits.controller.cleanup_timeout_ms	How long cleaning up leftovers may take	20000	ms	Editable / controller
controller.limits.controller.gateway_ready_timeout_ms	How long to wait for the gateway to become ready	90000	ms	Editable / controller
controller.limits.controller.gateway_probe_timeout_ms	How long one gateway readiness check may take	3000	ms	Editable / controller
controller.limits.controller.gateway_probe_interval_ms	How often the gateway is checked for readiness	1000	ms	Editable / controller
controller.limits.controller.shutdown_timeout_ms	How long CatalEx gets to stop cleanly	15000	ms	Editable / controller
controller.limits.controller.drain_timeout_ms	How long the gateway gets to finish work before stopping	20000	ms	Editable / controller
controller.limits.controller.control_timeout_ms	How long a request to the controller may take	5000	ms	Editable / controller
controller.limits.controller.down_timeout_ms	How long stopping CatalEx may take	25000	ms	Editable / controller
controller.limits.controller.down_poll_interval_ms	How often shutdown checks that services have stopped	250	ms	Editable / controller
controller.limits.controller.control_read_header_timeout_ms	How long a control client may take to send request headers	3000	ms	Editable / controller
controller.limits.controller.console_read_header_timeout_ms	How long a browser may take to send request headers	5000	ms	Editable / controller
controller.limits.controller.console_idle_timeout_ms	How long an idle browser connection is kept open	60000	ms	Editable / controller
controller.limits.controller.media_reconcile_interval_ms	How often stored media is checked against chats	3600000	ms	Editable / controller
controller.limits.controller.socket_probe_timeout_ms	How long checking a local socket may take	200	ms	Editable / controller
controller.limits.controller.orphan_stop_timeout_ms	How long to wait when stopping leftover processes	120000	ms	Editable / controller
controller.limits.controller.stats_interval_ms	How often resource usage is sampled	30000	ms	Editable / controller
controller.limits.controller.stats_docker_timeout_ms	How long reading container usage may take	5000	ms	Editable / controller
controller.limits.controller.stats_chats	How many chats are included in usage statistics	500	count	Editable / controller
controller.limits.controller.stats_processes	How many processes are included in usage statistics	256	count	Editable / controller
controller.limits.controller.slack_setup_timeout_ms	How long one Slack setup step on the messaging service may take	25000	ms	Editable / controller

Computer and account checks

Inspect network configuration, health-check cadence and cross-computer chat visibility. Use the supported connection/endpoint workflow for read-only addresses.

Key	Purpose	Default	Unit	Change / restart
controller.destination_check_interval_ms	How often connected accounts are checked for problems	60000	ms	Editable / none
controller.destination_check_timeout_ms	How long one account check may take	30000	ms	Editable / none
controller.port	The port CatalEx's main service listens on	7443	port	Read-only
controller.control_port	The port the dashboard uses to manage CatalEx	7449	port	Read-only
controller.native_chat_console_port	The port for the native chats' console	7445	port	Read-only
controller.native_chat_gateway_port	The port for the native chat gateway	7441	port	Read-only
controller.agent_console_port	The port for the agent console	7447	port	Read-only
controller.agent_gateway_port	The port agents use to reach the gateway	7442	port	Read-only
controller.service_port	The port of the service CatalEx connects to	8080	port	Read-only
controller.database_port	The port of CatalEx's database	15432	port	Read-only
controller.control_bind	The address the management service listens on	"127.0.0.1"	string	Read-only
controller.service_url	The address of the hosted service CatalEx connects to	""	string	Read-only
controller.show_other_computers_chats	Show chats from other computers in each account’s conversation list	false	boolean	Editable / none

Fixed protocol bounds

These are fixed contract values, not tuning controls. Split oversized content or use an attachment.

Key	Purpose	Default	Unit	Change / restart
fixed.message_text_characters	The longest message text, in characters, that CatalEx accepts	65536	characters	Read-only