sase monitor

Start, stop, list, and inspect monitored long-running commands

Reference for sase monitor. Run sase monitor --help for the exact output on this checkout.

usage: sase monitor [-h] {list,show,resume,start,stop} ...

Hand a slow command off to a detached supervisor that runs as a real agent session member, with streaming output, a timeout, and an optional follow-up agent. Running sase monitor defaults to sase monitor list.

Bare sase monitor defaults to sase monitor list.

Options:

Option	Description	Default
-h, --help	show this help message and exit	—

Subcommands:

Subcommand	Summary
sase monitor list	List monitors (rich table by default, -j/--json for JSON)
sase monitor resume	Resume a terminal monitor's requested follow-up without rerunning it
sase monitor show	Show one monitor and its captured output
sase monitor start	Start a command as a monitor agent session member
sase monitor stop	Stop a running monitor

sase monitor list

usage: sase monitor list [-h] [-a] [-f {table,markdown,json}] [-j] [-l NAME]
                         [-n N] [-p NAME] [-s STATE]

List monitor agent session members, newest first. By default this shows only active (running) monitors; pass --all to include finished ones too. Note: here -a is --all; unlike monitor start, use -l/--agent to filter by agent.

Options:

Option	Description	Default
-l NAME, --agent NAME	Only monitors belonging to this agent	—
-a, --all	Include finished monitors, not just running ones	false
-f FORMAT, --format FORMAT	Output format: 'table' (default), 'markdown', or 'json' (choices: json, markdown, table)	table
-h, --help	show this help message and exit	—
-j, --json	Emit a machine-readable JSON envelope (shorthand for --format json)	false
-n N, --limit N	Show at most N monitors	—
-p NAME, --project NAME	Only monitors from this project (default: every project)	—
-s STATE, --status STATE	Only monitors in this state; repeat to add more (running, completed, failed, timeout, stopped, lost) (choices: completed, failed, lost, running, stopped, timeout)	—

sase monitor resume

usage: sase monitor resume [-h] [-j] [-k FILE] [-m MODEL] ID

Resume the ordinary follow-up requested by a terminal monitor. This reconciles existing continuation delivery state first, then launches only from the frozen monitor result; it never reruns the monitored command.

Options:

Option	Description	Default
-k FILE, --checkpoint FILE	Optional authored checkpoint YAML/JSON for a manual recovery branch. Bound by content digest, not path	—
-h, --help	show this help message and exit	—
-j, --json	Emit a machine-readable JSON result	false
-m MODEL, --model MODEL	Model or alias for the resumed follow-up branch	—

Arguments:

Argument	Description
monitor_id	Monitor id (or unique prefix), member agent name, or owning agent name

sase monitor show

usage: sase monitor show [-h] [-A] [-b N] [-d] [-f {markdown,json}] [-F]
                         [-l N] [-o] [-R START:END]
                         ID

Show one monitor by id (or unique id prefix), member agent name, or owning agent name, followed by the tail of its captured output. --follow streams new output until the monitor reaches a terminal state, exactly like sase proc show --follow.

Options:

Option	Description	Default
-A, --all-lines	Print the whole retained output instead of a tail	false
-d, --diagnostics	Print bounded failed-stage diagnostics instead of raw output	false
-F, --follow	Stream new output until the monitor reaches a terminal state	false
-f FORMAT, --format FORMAT	Output format: 'markdown' (default) or 'json' (choices: json, markdown)	markdown
-h, --help	show this help message and exit	—
-l N, --log-lines N	Output lines to show (default: 200)	200
-b N, --max-bytes N	Maximum bytes for --diagnostics or --range output (default: 65536)	65536
-o, --output-only	Print only the captured output, with no surrounding detail	false
-R START:END, --range START:END	Print a bounded retained raw-output byte range	—

Arguments:

Argument	Description
monitor_id	Monitor id (or unique prefix), member agent name, or owning agent name

sase monitor start

usage: sase monitor start [-h] [-a NAME] [-f REF] [-C DIR] [-i DURATION]
                          [-J RUN] [-j] [-k FILE] [-L TEXT] [-m MODEL]
                          [-n TEXT] [-o MODE] [-P FILE] [-p NAME] [-r TEXT]
                          [-s TEXT] [-S TEXT] [-T N] [-t DURATION]
                          ...

Start a command under a detached supervisor and return. The supervisor owns the command's process group and streams its combined output, so the command outlives this process. When run inside an agent, this is the last output the agent produces before it is killed — the runner adopts the handoff and the monitor keeps running in the same agent and workspace.

Options:

Option	Description	Default
-a NAME, --agent NAME	Target agent (default: the calling agent). Note: unlike monitor list, here -a is --agent, not --all	—
-k FILE, --checkpoint FILE	Optional authored checkpoint YAML/JSON with objective, constraints, findings, unresolved decisions, remaining work, source refs, and coverage. Bound by content digest, not path	—
-f REF, --completion REF	Prepared host-completion intent reference from sase final prepare. Binding is atomic and single-use	—
-C DIR, --cwd DIR	Working directory (default: the current directory when it is inside a repo opened with sase repo open, else the agent's workspace, else $PWD)	—
-h, --help	show this help message and exit	—
-i DURATION, --idle-timeout DURATION	Optional no-output timeout; kills commands that stop producing bytes for this long, while quiet commands remain supported when omitted	—
-J RUN, --join RUN	Join a starter-scoped detached ToolRun instead of starting a command; cannot be used with a command, -c, -f, or -a	—
-j, --json	Emit the started monitor as a JSON envelope	false
-L TEXT, --label TEXT	Short row label (default: derived from the command)	—
-m MODEL, --model MODEL	Model or alias for the follow-up agent (opus, opus@high, sonnet, codex/gpt-5). Requires -n/--next. Default: inherit the starter's model	—
-n TEXT, --next TEXT	Next action for the follow-up agent; omit for no follow-up	—
-o MODE, --next-output MODE	How much retained output to hand the follow-up agent: 'auto' (default) selects outcome-aware evidence, 'tail' embeds the last --tail-lines lines fenced as untrusted data, 'file' points at refs and log locators instead, 'none' gives only the outcome summary and a sase monitor show --all-lines pointer (choices: auto, file, none, tail)	—
-P FILE, --policy FILE	Outcome-policy file. Mutually exclusive with -p/--profile	—
-p NAME, --profile NAME	Named outcome profile (verify). Profile selection alone never authorizes host completion (choices: verify)	—
-r TEXT, --reason TEXT	Why this command is being monitored (default: 'run command')	—
-s TEXT, --start-status TEXT	Label shown while the command runs (present tense, e.g. TESTING). Required unless -p/--profile supplies it. Pair with -S/--stop-status. Max 20 characters	—
-S TEXT, --stop-status TEXT	Label shown when the command finishes (past tense, e.g. TESTED). Required unless -p/--profile supplies it. Pair with -s/--start-status. Max 20 characters	—
-T N, --tail-lines N	Output tail handed to the follow-up (default: 200)	—
-t DURATION, --timeout DURATION	Timeout: bare seconds or a duration like 90s / 45m / 2h (default: 1h)	—

Arguments:

Argument	Description
monitor_command_words	The command to run and monitor, introduced by --

sase monitor stop

usage: sase monitor stop [-h] [-j] [-Q PATH] [-R PATH] [ID]

Stop a running monitor by id (or unique id prefix), member agent name, or owning agent name; omitted, targets the calling agent's active monitor. No follow-up agent is launched, even when a next action was recorded.

Options:

Option	Description	Default
-h, --help	show this help message and exit	—
-j, --json	Emit a machine-readable JSON result	false
-Q PATH, --request-path PATH	Private versioned operation request sidecar. Defaults to $SASE_PROC_REQUEST_PATH when that environment variable is set.	—
-R PATH, --result-path PATH	Private versioned typed result sidecar. Defaults to $SASE_PROC_RESULT_PATH when that environment variable is set.	—

Arguments:

Argument	Description
monitor_id	Monitor id (or unique prefix), member agent name, or owning agent name

Related: CLI command groups.

_[project]: A sase project is a named unit of work registered with SASE. _[workspace]: A sase workspace is a numbered clone of a project's primary repo, managed by the workspace store and tracked in that project's registry.json. _[agent]: A sase agent is an agent session or a single agent that does not belong to a session. _[prompt]: A prompt is the text SASE hands to a provider to start an agent turn. _[macro]: A macro is a reusable prompt triggered with #name in agent prompts. _[Patch]: A Patch is SASE's local unit of change. _[bead]: A bead is one tracked work item in SASE's git-native issue tracker. _[plan]: A plan is a persisted, approved artifact recording the intent behind agent work. _[TUI]: The TUI is SASE's terminal interface for running agents and reviewing results. _[epic]: An epic is an executable multi-phase plan run with sase bead work. _[gate]: A sase gate is a durable, command-backed request for one user decision, created when work must pause for the user. _[monitor]: A sase monitor is a monitor turn that runs one long command under a detached supervisor, so the command outlives the agent that started it. *[stitch]: A stitch is the lightweight ordered change record inside a Patch's STITCHES: section.