CLI reference¶
Generated
This page is generated from the CLI's own --help by
s/docs-cli-reference, so it cannot drift from the real command
surface. Do not edit it by hand. If it disagrees with your build,
trust your build.
Global options¶
Accepted by every command.
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
--format json changes only the result. Progress, warnings, and
errors always go to stderr as plain text, so piping stdout into jq is
safe even for commands that print several progress lines.
Commands¶
cortex session¶
Hold a persistent connection to the device, serving other commands [alias: connect].
Hold a persistent connection to the device, serving other commands.
The protocol requires one effective USB owner, although a damaging second open may succeed. This process claims ownership, performs the handshake ONCE, and serves every other command through local IPC.
That matters for more than speed. A held session SUBSCRIBES to device state, which is how the unit reports edits you make on the hardware - so what `cortex` reports can stay true while you play, rather than being a snapshot from whenever the last command ran.
`start` runs it in the background and returns once it is serving; `status` reports on it; `stop` ends it, announcing the disconnect to the device first.
Usage: cortex session [OPTIONS] <COMMAND>
Commands:
start Open the session and serve other commands
status Report whether a session is running, and whether the device answers
stop Ask a running session to shut down, announcing the disconnect first
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex session start¶
Open the session and serve other commands.
Open the session and serve other commands.
Runs in the BACKGROUND by default, detached from the terminal, so closing the terminal does not take the session with it. The log goes beside the socket in `$XDG_RUNTIME_DIR`.
It waits for the session to start serving before returning, so a handshake that fails is reported here rather than discovered by the next command.
Usage: cortex session start [OPTIONS]
Options:
--device <DEVICE>
Product whose USB interface the held daemon should own
[default: quad]
[possible values: quad, nano]
--foreground
Stay in the foreground, logging to the terminal.
This is what the background mode runs internally. Useful when a handshake is misbehaving and you want to watch it happen.
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex session start # background, detached
cortex session start --foreground # stay attached and watch it
cortex session status¶
Report whether a session is running, and whether the device answers.
Report whether a session is running, and whether the device answers
Usage: cortex session status [OPTIONS]
Options:
--device <DEVICE>
Product whose held session to inspect
[default: quad]
[possible values: quad, nano]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex session status
cortex session stop¶
Ask a running session to shut down, announcing the disconnect first.
Ask a running session to shut down, announcing the disconnect first
Usage: cortex session stop [OPTIONS]
Options:
--device <DEVICE>
Product whose held session to stop
[default: quad]
[possible values: quad, nano]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex session stop
cortex nano¶
Nano Cortex state and non-persistent working-state operations.
Nano Cortex state and non-persistent working-state operations
Usage: cortex nano [OPTIONS] <COMMAND>
Commands:
state Read the complete fixed eight-role signal-chain state
set-amp Set one amp control as raw 0-255 and verify through fresh read-back
set-gate-reduction Set Gate reduction as 0-100% and verify through fresh read-back
set-bypass Bypass or engage one Gate/FX role and verify through fresh read-back
read-fx-params Read named FX parameters (normalized 0.0-1.0) for one editable slot
set-fx-param Set one FX parameter (normalized 0.0-1.0) and verify through fresh read-back
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex nano state¶
Read the complete fixed eight-role signal-chain state.
Read the complete fixed eight-role signal-chain state
Usage: cortex nano state [OPTIONS]
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex nano set-amp¶
Set one amp control as raw 0-255 and verify through fresh read-back.
Set one amp control as raw 0-255 and verify through fresh read-back
Usage: cortex nano set-amp [OPTIONS] <CONTROL> <VALUE>
Arguments:
<CONTROL>
Amp control to change
[possible values: gain, level, bass, mid, treble]
<VALUE>
Raw device value from 0 to 255
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex nano set-gate-reduction¶
Set Gate reduction as 0-100% and verify through fresh read-back.
Set Gate reduction as 0-100% and verify through fresh read-back
Usage: cortex nano set-gate-reduction [OPTIONS] <PERCENT>
Arguments:
<PERCENT>
Gate reduction percentage from 0 to 100
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex nano set-bypass¶
Bypass or engage one Gate/FX role and verify through fresh read-back.
Bypass or engage one Gate/FX role and verify through fresh read-back
Usage: cortex nano set-bypass [OPTIONS] <TARGET> <BYPASSED>
Arguments:
<TARGET>
Gate or FX role to change
[possible values: gate, pre-fx1, pre-fx2, post-fx1, post-fx2, post-fx3]
<BYPASSED>
`true` to bypass, `false` to engage
[possible values: true, false]
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex nano read-fx-params¶
Read named FX parameters (normalized 0.0-1.0) for one editable slot.
Read named FX parameters (normalized 0.0-1.0) for one editable slot
Usage: cortex nano read-fx-params [OPTIONS] <SLOT>
Arguments:
<SLOT>
FX slot to read
[possible values: pre-fx1, pre-fx2, post-fx1, post-fx2, post-fx3]
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex nano set-fx-param¶
Set one FX parameter (normalized 0.0-1.0) and verify through fresh read-back.
Set one FX parameter (normalized 0.0-1.0) and verify through fresh read-back
Usage: cortex nano set-fx-param [OPTIONS] --expected-model-id <EXPECTED_MODEL_ID> <SLOT> <PARAM_INDEX> <VALUE>
Arguments:
<SLOT>
FX slot to change
[possible values: pre-fx1, pre-fx2, post-fx1, post-fx2, post-fx3]
<PARAM_INDEX>
Parameter index within the slot's model
<VALUE>
Normalized value, 0.0 to 1.0
Options:
--expected-model-id <EXPECTED_MODEL_ID>
Model id returned by `nano state` when the parameter was selected
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex preset¶
Presets: list, inspect, recall, prepare/save, or delete.
Presets: list, inspect, recall, prepare/save, or delete
Usage: cortex preset [OPTIONS] <COMMAND>
Commands:
copy Copy a stored preset through destination preparation, source recall, and save
delete Delete a preset from a setlist, by name
move Move a preset to an empty slot in the same setlist
prepare-save Prepare a save destination before editing the working grid
save Commit the working grid to a destination prepared before editing
list List the presets in a setlist, in slot order
show Recall a slot and dump the preset it loads
recall Recall a preset by slot, making it the one loaded on the grid
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex preset copy¶
Copy a stored preset through destination preparation, source recall, and save.
Copy a stored preset through destination preparation, source recall, and save.
WRITES TO THE UNIT and changes what is loaded. The destination is recalled and backed up before the source is recalled.
Usage: cortex preset copy [OPTIONS] --from <BANK+LETTER> --to <BANK+LETTER>
Options:
--from-setlist <PATH>
Source setlist path
[default: "/media/p4/Presets/My Presets"]
--from <BANK+LETTER>
Source slot
--to-setlist <PATH>
Destination setlist path
[default: "/media/p4/Presets/My Presets"]
--to <BANK+LETTER>
Destination slot
--name <NAME>
Destination name. Defaults to the recalled source preset's name
--instrument <INSTRUMENT>
Preferred-instrument metadata
[default: guitar]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex preset delete¶
Delete a preset from a setlist, by name.
Delete a preset from a setlist, by name.
WRITES TO THE UNIT, and there is no undo on the device.
Addressed by NAME, not slot - the opposite of `save`. Use the name the device reports in `cortex preset list`, which a save may have altered: on a name collision the unit de-duplicates with a `_N` suffix.
The factory library is refused.
Usage: cortex preset delete [OPTIONS] --name <NAME>
Options:
--name <NAME>
The preset's stored name, exactly as `cortex preset list` shows it
--setlist <PATH>
Absolute device path of the setlist
[default: "/media/p4/Presets/My Presets"]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex preset delete --name "SCRATCH"
cortex preset move¶
Move a preset to an empty slot in the same setlist.
Move a preset to an empty slot in the same setlist.
WRITES TO THE UNIT. The command requests a fresh complete listing and refuses an empty source, an occupied destination, a no-op move, the factory library, and malformed slots.
Usage: cortex preset move [OPTIONS] --from <BANK+LETTER> --to <BANK+LETTER>
Options:
--from <BANK+LETTER>
Occupied source slot: bank number then letter, e.g. `2A`
--to <BANK+LETTER>
Empty destination slot: bank number then letter, e.g. `2B`
--setlist <PATH>
Absolute device path of the setlist
[default: "/media/p4/Presets/My Presets"]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex preset list --include-empty
cortex preset move --from 2A --to 2B --dry-run
cortex preset move --from 2A --to 2B
cortex preset prepare-save¶
Prepare a save destination before editing the working grid.
Prepare a save destination before editing the working grid
Usage: cortex preset prepare-save [OPTIONS] --slot <BANK+LETTER>
Options:
--slot <BANK+LETTER>
Target slot: bank number then letter, e.g. `7A`
--setlist <PATH>
Absolute device path of the setlist
[default: "/media/p4/Presets/My Presets"]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex session start
cortex preset prepare-save --slot 7A
cortex preset save¶
Commit the working grid to a destination prepared before editing.
Commit the working grid to a destination prepared before editing.
WRITES TO THE UNIT, and there is no undo on the device. It overwrites whatever is in the slot.
What gets saved is the working grid - whatever `cortex grid show` reports - not a preset you name. Omit --name to keep the slot's existing name; give one to save into an empty slot or rename an occupied one.
The factory library is refused.
Usage: cortex preset save [OPTIONS] --token <TOKEN>
Options:
--token <TOKEN>
Opaque token returned by `preset prepare-save`
--name <NAME>
Name to save under. Omit to keep the slot's existing name
--instrument <INSTRUMENT>
Preferred-instrument metadata
[default: guitar]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex preset save --token save-1 --dry-run
cortex preset save --token save-1
cortex preset save --token save-1 --name "Lead Tone"
cortex preset list¶
List the presets in a setlist, in slot order.
List the presets in a setlist, in slot order.
Read-only: this does NOT change what is loaded on the grid.
Usage: cortex preset list [OPTIONS]
Options:
--setlist <PATH>
Absolute device path of the setlist, e.g. `/media/p4/Presets/My Presets`. Run `cortex setlist list` to list them
[default: "/media/p4/Presets/My Presets"]
--include-empty
Include empty slots, so a free slot can be found
--timeout <SECONDS>
Seconds to wait for the complete 256-slot listing. A timeout means no answer arrived, not that the setlist is empty
[default: 25]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex preset list
cortex preset list --include-empty
cortex preset show¶
Recall a slot and dump the preset it loads.
Recall a slot and dump the preset it loads.
CHANGES WHAT IS HEARD: there is no side-effect-free way to read a STORED preset - the device only emits a preset when it recalls one. Use `cortex grid show` if you want the live grid without recalling.
Usage: cortex preset show [OPTIONS] --slot <BANK+LETTER>
Options:
--slot <BANK+LETTER>
Slot name: bank number then letter, e.g. `1A`, `28C`. Bank is 1-32, letter A-H
--setlist <PATH>
Absolute device path of the setlist. `cortex setlist list` lists them
[default: "/media/p4/Presets/My Presets"]
--factory
Mark the setlist as the read-only factory library
--params
Also show each block's stored parameter values
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex preset show --slot 1B
cortex preset show --slot 1B --params
cortex preset recall¶
Recall a preset by slot, making it the one loaded on the grid.
Recall a preset by slot, making it the one loaded on the grid.
CHANGES WHAT IS HEARD. Nothing is saved and no stored preset is modified, but the grid is replaced and any unsaved edits are lost.
Usage: cortex preset recall [OPTIONS] --slot <BANK+LETTER>
Options:
--slot <BANK+LETTER>
Slot name: bank number then letter, e.g. `1A`, `12H`, `28C`. Bank is 1-32 and letter is A-H, giving 256 slots per setlist
--setlist <PATH>
Absolute device path of the setlist, e.g. `/media/p4/Presets/My Presets`. Run `cortex setlist list` to list them
[default: "/media/p4/Presets/My Presets"]
--factory
Mark the setlist as the read-only factory library. Needed for paths under /opt/neuraldsp/Factory Library
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex preset recall --slot 2B
cortex setlist¶
Setlists: the folders of presets the unit holds.
Setlists: the folders of presets the unit holds
Usage: cortex setlist [OPTIONS] <COMMAND>
Commands:
create Create a new USER setlist as a sibling of My Presets
delete Delete a USER setlist and all presets it contains
duplicate Duplicate a USER setlist through create plus recall/save per preset
list List every folder the device knows: setlists, captures, IR libraries
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex setlist create¶
Create a new USER setlist as a sibling of My Presets.
Create a new USER setlist as a sibling of My Presets
Usage: cortex setlist create [OPTIONS] --name <NAME>
Options:
--name <NAME>
Single setlist name, not a path
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex setlist delete¶
Delete a USER setlist and all presets it contains.
Delete a USER setlist and all presets it contains
Usage: cortex setlist delete [OPTIONS] --name <NAME>
Options:
--name <NAME>
Single setlist name, not a path. My Presets is always refused
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex setlist duplicate¶
Duplicate a USER setlist through create plus recall/save per preset.
Duplicate a USER setlist through create plus recall/save per preset
Usage: cortex setlist duplicate [OPTIONS] --source <NAME> --destination <NAME>
Options:
--source <NAME>
Source setlist name under the USER root
--destination <NAME>
New destination setlist name under the USER root
--limit <LIMIT>
Copy at most this many occupied presets
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex setlist list¶
List every folder the device knows: setlists, captures, IR libraries.
List every folder the device knows: setlists, captures, IR libraries.
A single `File` READ makes the device enumerate all its folders, which arrive over ten to twenty seconds. There is no total-count field on the wire, so this always waits the full window.
Read-only.
Usage: cortex setlist list [OPTIONS]
Options:
--window <SECONDS>
Seconds to gather folder announcements
[default: 20]
--show-empty
Also list folders holding no presets.
The unit reports hundreds of them, nearly all empty, which buries the two or three you actually use.
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex setlist list
cortex setlist list --show-empty
cortex grid¶
The signal grid that is loaded right now.
The signal grid that is loaded right now
Usage: cortex grid [OPTIONS] <COMMAND>
Commands:
show Show the LIVE grid: what is loaded right now, unsaved edits included
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex grid show¶
Show the LIVE grid: what is loaded right now, unsaved edits included.
Show the LIVE grid: what is loaded right now, unsaved edits included.
Read-only and side-effect free. This is the command to use while editing: `cortex preset show --slot X` reads a STORED slot and can only do so by recalling it, which discards unsaved edits and resets the active scene.
Usage: cortex grid show [OPTIONS]
Options:
--timeout <SECONDS>
Seconds to wait for the grid
[default: 15]
--params
Also show each block's stored parameter values, which is how to check an edit landed
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex grid show
cortex grid show --params
cortex block¶
One block in the grid: its parameters, bypass, model, or removal.
One block in the grid: its parameters, bypass, model, or removal
Usage: cortex block [OPTIONS] <COMMAND>
Commands:
param Set a block parameter on the grid
bypass Bypass or enable a block on the grid
set Place a model in a grid cell, creating or replacing a block
remove Remove the block at a grid cell
move Move one block to an empty grid cell and verify by live read-back
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex block param¶
Set a block parameter on the grid.
Set a block parameter on the grid.
CHANGES THE WORKING GRID. Nothing is saved: the edit lives on the grid until you save the preset or recall another, which discards it. A complete live-grid read must confirm the requested value before this command reports success.
Rows are given as the unit LABELS them, 1-4, not the zero-based wire index. Use `cortex grid show` to see what is where.
Usage: cortex block param [OPTIONS] --row <1-4> --column <0-7>
Options:
--row <1-4>
Grid row as shown on the unit, 1-4
--column <0-7>
Grid column, 0-7, left to right
--param <NAME>
Parameter by NAME, e.g. `GAIN`. The model is read from the cell and resolved through the device catalog. Safer than --index: indices are positional and not every one is a visible knob
--index <N>
Parameter by raw wire index. Prefer --param
--value <0.0-1.0>
Normalised value, 0.0-1.0, which is what the wire carries
--real <N>
Value in the parameter's OWN units (dB, ms, Hz). Converted using the catalog range, so it needs --param
--text <TEXT>
String value, for parameters that take one such as a cabinet's microphone selection
--scene <0-7>
Apply to this scene (0-7) rather than the active one. Sends three messages and LEAVES THE UNIT ON THAT SCENE
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex block param --row 1 --column 2 --param GAIN --real 7.5
cortex block param --row 1 --column 2 --index 0 --value 0.75
cortex block param --row 1 --column 2 --param GAIN --value 0.9 --scene 2
cortex block bypass¶
Bypass or enable a block on the grid.
Bypass or enable a block on the grid.
CHANGES THE WORKING GRID. Nothing is saved. Success requires a complete live-grid read confirming the active-scene bypass state.
Usage: cortex block bypass [OPTIONS] --row <1-4> --column <0-7>
Options:
--row <1-4>
Grid row as shown on the unit, 1-4
--column <0-7>
Grid column, 0-7
--bypass
Bypass the block. Omit to enable it
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex block bypass --row 1 --column 2 --bypass # bypass it
cortex block bypass --row 1 --column 2 # enable it
cortex block set¶
Place a model in a grid cell, creating or replacing a block.
Place a model in a grid cell, creating or replacing a block.
CHANGES THE WORKING GRID. Nothing is saved.
Verifies the device accepted it: a placement refused for want of DSP capacity is accepted on the wire and simply absent afterwards, with no error, so this waits for the device's echo and falls back to a live grid read-back when the echo is late.
Usage: cortex block set [OPTIONS] --row <1-4> --column <0-7> --model <ID>
Options:
--row <1-4>
Grid row as shown on the unit, 1-4
--column <0-7>
Grid column, 0-7
--model <ID>
Model by NUMERIC id. Find one with `cortex catalog --search`
--no-verify
Send without waiting for the echo. A DSP refusal then looks identical to success
--timeout <SECONDS>
Seconds to wait for the device's echo
[default: 5]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex block set --row 1 --column 2 --model 1001 # find ids with: cortex catalog --search
cortex block remove¶
Remove the block at a grid cell.
Remove the block at a grid cell.
CHANGES THE WORKING GRID. Nothing is saved. Success requires a complete live-grid read confirming that the cell is empty.
Usage: cortex block remove [OPTIONS] --row <1-4> --column <0-7>
Options:
--row <1-4>
Grid row as shown on the unit, 1-4
--column <0-7>
Grid column, 0-7
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex block remove --row 1 --column 2
cortex block move¶
Move one block to an empty grid cell and verify by live read-back.
Move one block to an empty grid cell and verify by live read-back.
CHANGES THE WORKING GRID. Nothing is saved. A cross-row move lets the device create or adjust a parallel path.
Usage: cortex block move [OPTIONS] --from-row <1-4> --from-column <0-7> --to-row <1-4> --to-column <0-7>
Options:
--from-row <1-4>
Source row as shown on the unit, 1-4
--from-column <0-7>
Source column, 0-7
--to-row <1-4>
Destination row as shown on the unit, 1-4
--to-column <0-7>
Empty destination column, 0-7
--timeout <SECONDS>
Seconds to wait for each live-grid read
[default: 15]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex block move --from-row 1 --from-column 2 --to-row 2 --to-column 6
cortex row¶
A grid row: its input, its output, and where it splits.
A grid row: its input, its output, and where it splits
Usage: cortex row [OPTIONS] <COMMAND>
Commands:
input Re-point a grid row's input
output Re-point a grid row's output
split Set a row's split and mix points, activating a parallel branch
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex row input¶
Re-point a grid row's input.
Re-point a grid row's input.
CHANGES THE WORKING GRID. Nothing is saved. Success requires a complete live-grid read confirming the typed route.
Usage: cortex row input [OPTIONS] --row <1-4> --port <PORT>
Options:
--row <1-4>
Grid row as shown on the unit, 1-4
--port <PORT>
Typed input destination, e.g. input1, return1, usb5, previous_row
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex row input --row 1 --port input1
cortex row output¶
Re-point a grid row's output.
Re-point a grid row's output.
CHANGES THE WORKING GRID. Nothing is saved. Success requires a complete live-grid read confirming the typed route.
Not every destination is physical: next_row3, next_row4 and next_row34 route internally, while multiple is a real device-selected output.
Usage: cortex row output [OPTIONS] --row <1-4> --port <PORT>
Options:
--row <1-4>
Grid row as shown on the unit, 1-4
--port <PORT>
Typed output destination, e.g. xlr12, out3, usb5, next_row3
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex row output --row 1 --port xlr12
cortex row split¶
Set a row's split and mix points, activating a parallel branch.
Set a row's split and mix points, activating a parallel branch.
CHANGES THE WORKING GRID. Nothing is saved. Success requires a complete live-grid read confirming both control points.
Only screen rows 1 and 3 can branch; their parallel lane is the row below. Rows 2 and 4 have no splitter and are refused.
Usage: cortex row split [OPTIONS] --row <1|3> --split <COL>
Options:
--row <1|3>
Grid row as shown on the unit. Must be 1 or 3
--split <COL>
Column at which the row branches. -1 clears the branch
--mix <COL>
Column at which the branch rejoins. -1 means it never does
[default: -1]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex row split --row 1 --split 3 --mix 6 # branch at 3, rejoin at 6
cortex row split --row 1 --split -1 --mix -1 # clear the branch
cortex device¶
The unit itself: firmware, DSP load, and a connection probe.
The unit itself: firmware, DSP load, and a connection probe
Usage: cortex device [OPTIONS] <COMMAND>
Commands:
version Read the device firmware version (CorOS, app, bootloader, zencoder)
cpu Show the unit's live DSP load
probe Probe the core session read paths and report their state
help Print this message or the help of the given subcommand(s)
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex device version¶
Read the device firmware version (CorOS, app, bootloader, zencoder).
Read the device firmware version (CorOS, app, bootloader, zencoder)
Usage: cortex device version [OPTIONS]
Options:
--session
Read via the session layer (background RX thread + correlated request) instead of the one-shot synchronous transport.
Both paths are valid: a `Version` READ is answered without the connect handshake. This flag exists to exercise the session layer against hardware. Both paths are now verified; see spec/140-session/spec.md.
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex device version
cortex device version --format json
cortex device cpu¶
Show the unit's live DSP load.
Show the unit's live DSP load.
The device pushes this about once a second, but only to a client that has subscribed - so this needs a running `cortex session start`. A one-shot command uses a minimal handshake and never asks the device to push it.
Usage: cortex device cpu [OPTIONS]
Options:
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex device cpu # needs a running: cortex session start
cortex device probe¶
Probe the core session read paths and report their state.
Probe the core session read paths and report their state.
Without a held daemon this performs a subscribed handshake, reads the core scene/current-preset/preset-list paths, then disconnects. With a held daemon it probes those paths through the existing owner and leaves that session running.
Read-only: the handshake sends READs and a connect announcement. It never writes preset data and never saves.
Usage: cortex device probe [OPTIONS]
Options:
--listen <SECONDS>
Extra seconds to hold the session open after the handshake before reading. The handshake already waits for the device to go quiet, so 0 is usually fine; raise it if reads time out
[default: 0]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex device probe
cortex setup¶
Diagnose local installation, USB access, daemon health, and MCP setup.
Diagnose local installation, USB access, daemon health, and MCP setup.
This command never opens the device. By default it only reports what needs attention. The two changes it can make require explicit flags.
Usage: cortex setup [OPTIONS]
Options:
--install-udev
Install or replace the narrowly-scoped udev rule using sudo, then reload and trigger the rules. Replug the device afterwards
--claude-code
Register the sibling cortex-mcp binary with Claude Code at user scope. This changes Claude Code configuration, not device state
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex setup
cortex setup --install-udev
cortex setup --claude-code
cortex scene¶
Switch, label, recolour, copy, or swap scenes.
Switch, label, recolour, copy, or swap scenes.
CHANGES WHAT IS HEARD. Nothing is saved.
Usage: cortex scene [OPTIONS] [COMMAND]
Commands:
switch Switch the active scene. Changes what is heard; does not save
label Set a scene label on the unsaved working copy
unlabel Clear a scene label on the unsaved working copy
color Set a scene colour as `0xAARRGGBB`, `#RRGGBB`, or decimal
copy Copy one scene onto another, including its label and colour
swap Exchange two scenes, including their labels and colours
help Print this message or the help of the given subcommand(s)
Options:
--index <0-7>
Scene number, 0-7 ZERO-BASED: 0 is scene A and 7 is scene H. The unit labels them A-H, so scene C is `--index 2`. Compatibility shorthand for `cortex scene switch --index N`
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex scene switch --index 2
cortex scene label --index 2 --label 'Wide Lead'
cortex scene color --index 2 --color '#FF02C2'
cortex scene copy --from 1 --to 3
cortex scene swap --first 1 --second 3
Compatibility:
cortex scene --index 2
cortex scene switch¶
Switch the active scene. Changes what is heard; does not save.
Switch the active scene. Changes what is heard; does not save
Usage: cortex scene switch [OPTIONS] --index <0-7>
Options:
--index <0-7>
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex scene label¶
Set a scene label on the unsaved working copy.
Set a scene label on the unsaved working copy
Usage: cortex scene label [OPTIONS] --index <0-7> --label <LABEL>
Options:
--index <0-7>
--label <LABEL>
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex scene unlabel¶
Clear a scene label on the unsaved working copy.
Clear a scene label on the unsaved working copy
Usage: cortex scene unlabel [OPTIONS] --index <0-7>
Options:
--index <0-7>
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex scene color¶
Set a scene colour as 0xAARRGGBB, #RRGGBB, or decimal.
Set a scene colour as `0xAARRGGBB`, `#RRGGBB`, or decimal
Usage: cortex scene color [OPTIONS] --index <0-7> --color <COLOR>
Options:
--index <0-7>
--color <COLOR>
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex scene copy¶
Copy one scene onto another, including its label and colour.
Copy one scene onto another, including its label and colour
Usage: cortex scene copy [OPTIONS] --from <0-7> --to <0-7>
Options:
--from <0-7>
--to <0-7>
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex scene swap¶
Exchange two scenes, including their labels and colours.
Exchange two scenes, including their labels and colours
Usage: cortex scene swap [OPTIONS] --first <0-7> --second <0-7>
Options:
--first <0-7>
--second <0-7>
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
cortex catalog¶
Search or inspect the device model catalog, or save its raw payload.
Search or inspect the device model catalog, or save its raw payload.
The catalog is what turns the integer model ids stored in a preset into names. It comes from the device, so it reflects installed model content rather than a hard-coded factory table.
Read-only.
Usage: cortex catalog [OPTIONS]
Options:
--search <TEXT>
Case-insensitive substring to match against a model's name AND the gear it is based on, e.g. `marshall`, `tape echo`
--model <ID>
Show one model in full by NUMERIC id, e.g. `--model 1001`. Use --search to find an id from a name
--dump <FILE>
File path to write the raw payload to, for inspection
--from-file <FILE>
Parse a payload previously saved with --dump, instead of asking the device. Needs no unit connected
--timeout <SECONDS>
Seconds to wait. The payload is ~47 KB over several hundred reports, so allow generously
[default: 40]
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex catalog --search plexi
cortex catalog --model 1001
cortex catalog --dump repo.bin
cortex decode-trace¶
Decode a USB capture into Cortex Control messages.
Decode a USB capture into Cortex Control messages.
Reads `tshark` field output on standard input and prints one line per reassembled message, in the same shape as `CORTEX_TRACE`. Use `s/usb-decode`, which supplies the right `tshark` invocation.
Works on a capture of ANY client, which is the point: it is how a recording of Cortex Control gets read against our own schema.
Usage: cortex decode-trace [OPTIONS]
Options:
--quiet
Print only the summary line, not every message
--verbose
Also describe each message's protobuf fields.
Generic, so it works on message types we do not model - which is the case that matters when reading a capture of another client.
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
s/usb-decode traces/capture.pcapng # the usual way
s/usb-decode --live # watch it happen
cortex completions¶
Generate or install shell completions.
Generate or install shell completions.
`cortex completions install` is the one to use: it detects your shell, writes the completion file to the standard location, and prints any one-time setup still needed. It never edits your shell startup files.
`cortex completions <shell>` prints the script to stdout instead, which is the stable interface for packagers and unusual setups.
Usage: cortex completions [OPTIONS] <TARGET>
Arguments:
<TARGET>
A shell to generate for, or `install` to install for your own
Possible values:
- install: Detect the current shell and install into its standard directory
- bash
- elvish
- fish
- powershell
- zsh
Options:
--shell <SHELL>
Override the shell detected from $SHELL when using `install`
[possible values: bash, elvish, fish, powershell, zsh]
--dir <DIR>
Directory to write the correctly-named file into, instead of stdout (or instead of the standard location, with `install`)
--format <FORMAT>
Output format. `text` is for humans; `json` is for scripts and agents, and is stable enough to parse.
Only the RESULT changes format. Progress, warnings, and errors always go to stderr as plain text, so `cortex preset list --format json | jq` gets clean JSON regardless.
Possible values:
- text: Human-readable, the default
- json: Machine-readable JSON
[default: text]
--schema
Print the shared agent-operation JSON Schemas used by cortex-mcp
--zero-based
Take `--row` as 0-3 rather than the 1-4 shown on the unit.
The unit labels its rows 1-4 and the wire numbers them 0-3, so the default matches what a player sees. Scripts and agents generally have a zero-based index already, and converting it back by hand is exactly the sort of arithmetic that silently edits the wrong row.
-n, --dry-run
Print the operation plan without changing device or local state. Read-only commands accept and ignore this flag
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
Examples:
cortex completions install