mxr draft
Generated from
mxr draft --help. After editingcrates/daemon/src/cli/, update and review the snapshots withscripts/cargo-test -p mxr --test integration cli_help::, then rebuildsite/.
Draft a new email or refine an existing local draft with LLM assistance
mxr draft [OPTIONS] [COMMAND]Subcommands
Section titled “Subcommands”| Subcommand | Purpose |
|---|---|
refine |
Refine an existing local draft |
eval |
Measure drafting with your LLM. Nothing is saved or sent |
help |
Print this message or the help of the given subcommand(s) |
Options
Section titled “Options”| Flag | Description |
|---|---|
--to <TO> |
Recipient email for a new draft |
--purpose <PURPOSE> |
Plain-language purpose for a new draft |
--account <ACCOUNT> |
Account key, email, or id |
--register <REGISTER> |
Register to use when recipient has no relationship profile [possible values: casual, neutral, formal] |
--length <LENGTH> |
Desired length [possible values: short, medium, long] |
--format <FORMAT> |
[possible values: table, json, jsonl, csv, ids] |
-h, --help |
Print help |
Use When
Section titled “Use When”Get a suggested revision of an existing local draft from your configured LLM, or measure drafting with eval. refine prints the suggestion without replacing the draft or sending it; eval saves and sends nothing. See check how close drafts get.
Everyday Examples
Section titled “Everyday Examples”mxr draft refine DRAFT_ID --shortermxr draft refine DRAFT_ID --add-context 'The deadline is Friday'mxr draft eval --limit 20mxr draft eval --set quickSub-subcommands
Section titled “Sub-subcommands”mxr draft eval
Section titled “mxr draft eval”Measure drafting with your LLM. Nothing is saved or sent.
With –set mine (the default), replay your most recent replies: draft each one from what was known just before you replied and compare it with what you sent (greeting, sign-off, length, invented numbers). With –set quick, draft each labelled compose case and grade it against what that case required (kept sentences, answers, gaps, invented numbers, length, leftovers). Calls your drafting LLM once per case. –rewrite also runs the configured humanizer rewrite stage for each replay. Uses –account with –set mine.
mxr draft eval [OPTIONS]Options
Section titled “Options”| Flag | Description |
|---|---|
--limit <LIMIT> |
How many replies to replay, at most 50 (–set mine only) [default: 10] |
--set <SET> |
Which cases to draft Possible values: - mine: Your most recent replies, replayed - quick: The labelled compose corpus, graded check by check [default: mine] |
--rewrite |
Allow the configured humanizer rewrite stage for each replay. Runs only when humanizer and auto-fix are enabled and the draft scores below the threshold. Skipped rewrites are not marked |
-h, --help |
Print help (see a summary with ‘-h’) |
mxr draft refine
Section titled “mxr draft refine”Refine an existing local draft
mxr draft refine [OPTIONS] <DRAFT_ID>Arguments
Section titled “Arguments”| Flag | Description |
|---|---|
<DRAFT_ID> |
Options
Section titled “Options”| Flag | Description |
|---|---|
--shorter |
|
--warmer |
|
--more-formal |
|
--less-emoji |
|
--add-context <ADD_CONTEXT> |
|
-h, --help |
Print help |
See also
Section titled “See also”- CLI overview: full command index
- Concepts: query operators, search modes, JSON shapes
- Automation contract: which commands support
--format json,--dry-run, stdin