Commands

usage: ade budget [-h] [--refusals]

print core lines, hooks and refusal codes against their budgets; fails when one is over;
--refusals counts firings

options:
  -h, --help  show this help message and exit
  --refusals  each refusal code: times fired, and on cards that landed after

usage: ade card [-h] {add,list,show} ...

add, list and show cards; a card needs paths and a done line

positional arguments:
  {add,list,show}
    add            add a card
    list           list cards
    show           show one card with its latest stage, review and test

options:
  -h, --help       show this help message and exit

usage: ade cockpit [-h] [--mode {cli,orca,auto}]

the pilot's view: every open card with its state and next command, plus budgets; orca mode opens
it in Orca too

options:
  -h, --help            show this help message and exit
  --mode {cli,orca,auto}
                        orca also opens the board in an Orca terminal; auto does when Orca is
                        running; default [cockpit] mode

usage: ade conduct [-h] [--lanes N] [--json] [--follow] [--rounds N] [--check] [cards ...]

walk the board through `ade loop` until no open card is left untried, one card at a time or
several whose reach the graph keeps apart; route every stop by ade/route.toml, wake a model only
where the route says wake

positional arguments:
  cards       cards to conduct, in order; none means every open card, oldest first

options:
  -h, --help  show this help message and exit
  --lanes N   fly up to N cards at once, each in its own session, picking only cards whose reach
              in the graph does not meet a card already flying
  --json      after each parked card, one JSON line like `ade loop --json` (not with --lanes)
  --follow    after a run ends, sleep ADE_FOLLOW_POLL seconds (default 60) and run again, so a
              card added meanwhile is flown; stops on a provider over its week share or a new day
              (not with named cards)
  --rounds N  with --follow, stop after N rounds
  --check     only check the route and bound scripts against the verbs

usage: ade decide [-h] [--path PATH] [--ask] [--rec REC] [--default DEFAULT] [--by BY]
                  [--answer ID] [--list] [--queue]
                  [text]

rulings that govern path globs, and questions for a person with a recommendation, a default and a
deadline

positional arguments:
  text

options:
  -h, --help         show this help message and exit
  --path PATH        glob the decision governs (repeatable)
  --ask              record TEXT as a question for a person, not a ruling
  --rec REC          the question's recommended answer
  --default DEFAULT  the answer that applies if nobody answers by the deadline
  --by BY            deadline: YYYY-MM-DD, or Nd for N days from today
  --answer ID        answer question ID with TEXT
  --list             every ruling and question
  --queue            open questions, nearest deadline first

usage: ade digest [-h] [--day DAY]

one short summary of a day: cards landed, cards parked with their open question, questions whose
default applied

options:
  -h, --help  show this help message and exit
  --day DAY   the day to summarise, YYYY-MM-DD (default today)

usage: ade graph [-h] [--check]

regenerate the graph file from headers and imports; --check fails if the committed one is stale or
the route has drifted

options:
  -h, --help  show this help message and exit
  --check     fail if the graph file does not match the code

usage: ade handover [-h] [--role ROLE]

generate a role's working context from records and graph; prints its token estimate, fails over
budget

options:
  -h, --help   show this help message and exit
  --role ROLE

usage: ade hook [-h] {brief,bash,grep,turn-end,start}

Claude Code hook entry: brief (first-edit briefing, protected paths), bash (the guard), grep
(graph notes on search output), turn-end (holds), start (handover)

positional arguments:
  {brief,bash,grep,turn-end,start}

options:
  -h, --help            show this help message and exit

usage: ade init [-h] [--dry-run] [--starter {python,typescript}] [--yes] [--no-commit]

start a new project in an empty folder, or install or refresh ADE in a repo from ade.toml;
idempotent, --dry-run prints the diff

options:
  -h, --help            show this help message and exit
  --dry-run             print what would change, write nothing
  --starter {python,typescript}
                        for an empty folder: the starter project to write (default python)
  --yes                 do not open the setup picker on a first ade.toml
  --no-commit           in an existing repo, leave the written files uncommitted

usage: ade land [-h] card

land a card whose lane tree has an approved review and passing tests: merge, verdict, outcome

positional arguments:
  card

options:
  -h, --help  show this help message and exit

usage: ade link [-h] [--dry-run]

make a plain `ade` work in new shells: one small launcher named ade in a per-user folder already
on PATH

options:
  -h, --help  show this help message and exit
  --dry-run   print what would be written, write nothing

usage: ade loop [-h] [--json] card

drive a card through lane, build, review, test, land; journal each stage, stop at the first non-
advance

positional arguments:
  card

options:
  -h, --help  show this help message and exit
  --json      end the stop with one JSON line: card, stage, outcome, next

usage: ade observe [-h] [--recheck RECHECK] [--update ID] [--list] [MECHANISM STATE ...]

dated observations of how a mechanism actually behaves, each with a recheck date so none goes
stale unseen

positional arguments:
  MECHANISM STATE

options:
  -h, --help         show this help message and exit
  --recheck RECHECK  when to look again: YYYY-MM-DD, or Nd (default 30d)
  --update ID        record a fresh STATE for observation ID
  --list

usage: ade parts [-h] [--why WHY] [{set}] [ID=off|on]

list the parts catalogue with each part's state; `set x=off` (or `x=on`) switches one in ade.toml
and prints what it costs

positional arguments:
  {set}       set: switch a part
  ID=off|on   the part and its new state

options:
  -h, --help  show this help message and exit
  --why WHY   the reason you go past a warning, kept in [parts.override]

usage: ade push [-h] [--dry-run]

push trunk, ade-records and checkpoint tags to the remote; fast-forward only, never forced; --dry-
run asks the remote without pushing

options:
  -h, --help  show this help message and exit
  --dry-run   show what would be pushed; push nothing

usage: ade q [-h] {run,show,grep,sql,list} ...

keep long output out of a model's context: run a command through the store, page or grep what it
kept, query it

positional arguments:
  {run,show,grep,sql,list}
    run                 run a command; long output is stored and shown as head, tail and a handle
    show                one page of a stored output
    grep                lines matching a regex in one stored output, or in all of them
    sql                 a read-only query over outputs, records, files and edges
    list                the newest stored outputs

options:
  -h, --help            show this help message and exit

usage: ade rehearse [-h]
                    [--only {cockpit-orca-absent,graph-off,install-existing,newcomer-python,newcomer-typescript,per-file-tests,prose-only-review-none,review-off}]
                    [--keep]

one full loop per config option on throwaway repos with the fake provider; reports calls, wall
time and refusals

options:
  -h, --help            show this help message and exit
  --only {cockpit-orca-absent,graph-off,install-existing,newcomer-python,newcomer-typescript,per-file-tests,prose-only-review-none,review-off}
                        run just this scenario (repeatable)
  --keep                keep the scenario folders and print where they are

usage: ade session [-h] {start,list,attach,stop} ...

run an ade verb detached from this terminal (no tmux, any platform), list sessions live or ended,
follow one, stop one

positional arguments:
  {start,list,attach,stop}
    start               run one ade verb detached, its output in .ade/sessions/NAME.log
    list                every session, live or ended
    attach              follow a session's log until it ends (Ctrl-C leaves it running)
    stop                end a session

options:
  -h, --help            show this help message and exit

usage: ade setup [-h]

a terminal panel to switch the kit's parts on and off; shows the ade.toml diff first, saves
through parts.rewrite; a numbered list when there is no terminal or NO_COLOR is set

options:
  -h, --help  show this help message and exit

usage: ade status [-h] [--line]

one-line state for the Claude Code status line: repo, open cards, last stage outcome

options:
  -h, --help  show this help message and exit
  --line      status-line form; never fails

usage: ade telemetry [-h] [{export}]

show usage per card and per provider for the week, each week meter against its share; export OTLP
files locally

positional arguments:
  {export}    write OpenTelemetry JSON to .ade/telemetry/

options:
  -h, --help  show this help message and exit

usage: ade uninstall [-h] [--dry-run]

remove everything `ade init` generated (hooks, rules, agents, AGENTS.md block, graph); keep
ade.toml and records

options:
  -h, --help  show this help message and exit
  --dry-run   print what would change, remove nothing

usage: ade wait [-h] [--bound BOUND] [--lane CARD] [--poll POLL]
                {answered,clean,exit,file,journal,time} [arg]

wait for a condition with a bound (exit 0 met, 4 bound passed, 5 lane stalled); run it in the
background to be told

positional arguments:
  {answered,clean,exit,file,journal,time}
  arg                   the card, decision id, pid, path or duration the condition watches

options:
  -h, --help            show this help message and exit
  --bound BOUND         give up after this long (90s, 30m, 2h); every wait has one
  --lane CARD           also stop if this card's lane log goes quiet for [watch] stall minutes
  --poll POLL           seconds between looks

usage: ade watch [-h] [--once] [--until-landed] [--poll POLL] card

follow a card's lane log live (any platform, no tmux) until its loop stops or the log goes quiet;
--once prints it

positional arguments:
  card

options:
  -h, --help      show this help message and exit
  --once          print the log so far and exit
  --until-landed  keep following across stops until the card lands
  --poll POLL     seconds between looks