CLI command reference¶
This page is the canonical command inventory for the Crucible CLI. Use crucible <command> --help or crucible <resource> <action> --help for the exact arguments, aliases, defaults, and examples supported by the installed version.
Project options use --project-id with the conventional -p short alias. Sample type options use --type with -t. Older spellings were removed in 4.0.0 and stop with the option to use instead.
Global options¶
| Option | Description |
|---|---|
--version |
Print the installed client version and exit |
--debug |
Enable Crucible debug logging; place it before the command |
-E, --elevate |
Run one command with platform-administrator elevation; place it before the command |
--no-color |
Disable ANSI colors while retaining interactive terminal hyperlinks; place it before the command |
--no-pager |
Print a long listing directly even when the pager setting is auto; place it before the command |
By default no privilege mode is sent and the API decides: a platform administrator using an API key runs elevated and sees every record, a signed-in administrator runs normal, and everyone else gets their normal, ACL-derived access. whoami shows which mode applies. Set crucible config set privilege_mode normal to restrict an administrator account to its own access; --elevate (-E) then widens a single command. A future release will default to normal for API keys too. Administrator-only commands, such as service-account show, service-account set-role, and deletion review, elevate on their own and do not need the flag.
Explicit elevation can also come from the privilege_mode config key or the CRUCIBLE_PRIVILEGE_MODE environment variable. Because neither is visible on the command line, the CLI prints a notice to stderr when a command runs elevated without --elevate. In the interactive shell, elevate on|off toggles it for the session, elevate COMMAND runs a single command elevated, and an ELEVATED badge appears in the toolbar while the toggle is on. The shell switches elevation off after elevation_timeout idle minutes (default 15, 0 never).
deletion list shows the caller's review queue by default when they can review requests. Pass --scope to choose a different view.
Commands that create projects, instruments, or service accounts check the caller's account capabilities first and refuse immediately rather than prompting for values the API would reject.
Running crucible without a command starts the interactive shell. See the CLI overview for setup, shell completion, and interactive usage.
Listing¶
- List commands show 50 records by default (the
default_limitconfig key).--limit Nchanges that for one command, and--allfetches every match. - The header shows how many matched, for example
Projects (50 of 6,770). When records were left out, a note below the table says how many and how to see them. - With
--jsonor piped output, that note goes to stderr, so stdout stays valid data. - In the interactive shell,
moreshows the next page of the last listing. - Search commands keep their own
--limit, capped by the server. - Listings print directly.
crucible config set pager autoopens long listings in$CRUCIBLE_PAGER,$PAGER, orless -FRX;--no-pagerskips it for one command.
Errors¶
- A mistyped command or flag ends with a "Did you mean ...?" suggestion.
- API errors add a next step where one applies:
crucible loginorcrucible config initfor a rejected key or sign-in,crucible whoamifor missing permissions,read_timeoutfor slow responses. - With
--json, the hint appears aserror.hint.
--json is available for dataset, sample, project, and instrument list, get, search, and search-metadata; dataset list-parents, list-children, list-samples, list-files, list-keywords, and facets; sample list-parents, list-children, list-datasets, and facets; project and instrument list-users; user get, list, search, list-datasets, list-projects, list-access-groups, list-keys, approve, and set-role; service-account get, show, list, set-role, and list-access-groups; file, ingestion, ag, and deletion list and get; ag mine, deletion list-deleted and get-deleted, file ingestion, and ingestion list-ingestors; account show, keys list, and keys create; print barcode; and whoami and status. List and search commands return arrays, while single-record commands, facets, whoami, and status return objects. dataset list-files adds each file's signed download_url. The generic get command uses -o json (--output json) instead.
Dataset commands¶
| Command | Description |
|---|---|
dataset list |
List datasets oldest to newest, in the current project or across every accessible project, with project, instrument, linked-sample, metadata, empty-field, and name-pattern filters |
dataset get MFID |
Show a dataset, its files, and linked resources |
dataset create [--input FILE ...] |
Create a dataset, optionally uploading or cataloging files |
dataset update MFID |
Update model fields or scientific metadata |
dataset edit MFID |
Edit dataset fields interactively |
dataset reassign-project MFID PROJECT |
Move a dataset to another project; PROJECT accepts the project ID or MFID |
dataset transfer-ownership MFID USER |
Transfer dataset ownership |
dataset delete MFID |
Permanently delete a dataset after confirmation |
dataset facets FIELD |
Group datasets into value buckets with counts |
dataset search QUERY |
Search dataset names |
dataset search-metadata QUERY |
Search scientific metadata; search-md is an alias |
dataset link |
Link parent and child datasets; --relationship-type records the kind of link |
dataset unlink -p PARENT -c CHILD |
Remove a dataset parent-child link |
dataset list-parents MFID |
List parent datasets; --relationship-type filters by kind of link |
dataset list-children MFID |
List child datasets; --relationship-type filters by kind of link |
dataset link-sample MFID -s SAMPLE |
Link a sample to a dataset |
dataset unlink-sample MFID -s SAMPLE |
Unlink a sample from a dataset |
dataset list-samples MFID |
List samples linked to a dataset |
dataset add-file MFID -i FILE [FILE ...] |
Upload files to an existing dataset |
dataset add-thumbnail MFID IMAGE |
Encode and add a local image as a dataset thumbnail |
dataset list-files MFID |
List associated files and available download links |
dataset download MFID |
Download dataset files with optional include and exclude patterns |
dataset ingestion MFID |
Show ingestion requests for a dataset |
dataset add-keyword MFID WORD |
Add a keyword |
dataset list-keywords MFID |
List a dataset's keywords and usage counts |
dataset access ... |
List, grant, or revoke direct access entries |
dataset set-public MFID |
Make a dataset publicly viewable |
dataset set-private MFID |
Remove public access from a dataset |
dataset parsers |
List installed client-side parsers |
dataset pipelines |
List postprocessing pipelines and their parameters |
dataset postprocess MFID PIPELINE |
Queue a postprocessing run; --param KEY=VALUE (repeatable), --params JSON, --idempotency-key |
dataset list-runs [MFID] |
List postprocessing runs; --pipeline, --status |
dataset get-run ID |
Show a postprocessing run |
Common creation example:
crucible dataset create -i data.csv --project-id my-project \
-n "XRD measurement" -m "X-ray diffraction" \
--metadata '{"temperature_K": 300}' --keywords "XRD,powder"
Create a dataset record without attaching files:
Dataset creation accepts --project-id or --project-mfid and --instrument-id or --instrument-mfid. Matching ID and MFID forms may be supplied together. Interactive use remains ID-oriented, while MFID flags support integrations and automation.
Creating records and uploading files¶
sample createanddataset createaccept--owner USER(username, ORCID, MFID, or email). Platform contributors and administrators can name anyone; everyone else only themselves.- Run on a terminal without
--inputor--name,dataset createprompts for the name, project (default: the current project), measurement, session, and timestamp. Scripts are not prompted. --type,--ingestor,--no-upload,--backend,--access-note,--replace, and--no-ingestrequire at least one--inputfile.- Uploads use signed multipart uploads.
- On a terminal,
dataset createanddataset add-fileshow a progress line per file. --no-ingestuploads without requesting ingestion; request it later withcrucible file request-ingestion FILE_MFID. It cannot be combined with--waitor--ingestor.- When a file exists with different content, both commands ask whether to replace it on a terminal.
--replacereplaces without asking; scripts without it fail. --ingestorcompletes from the ingestion classes the API offers, cached for a day.
Use dataset list --project-id PROJECT --project-scope shared to show resources shared with a project but assigned elsewhere or unassigned. Use --project-scope all to combine assigned and shared resources. --project-mfid accepts the canonical project MFID instead of a project ID. The interactive shell completes both identifiers through project search. Human-readable scoped results include the resource's actual project and its assigned or shared relation.
dataset list and sample list use the current project when one is set and otherwise list every accessible project; --all-projects ignores the current project, and the table then shows each record's project. Results are fetched newest first, so --limit N keeps the N most recent, and printed oldest to newest so the newest row sits next to the prompt. --sort created|updated|name with --direction asc|desc changes the server ordering (--direction requires --sort); an ascending sort prints in that order. Tables are flat unless --group-by or the dataset_group_by/sample_group_by config key selects a field; groups keep chronological order and the group with the newest record comes last. --json returns records in server order.
Both commands accept --visibility all|public|private, --mine (records you own), --owner USER (username, ORCID, MFID, or email, resolved to the stable identifier before the request; repeat it for any of up to 10 owners), the time bounds --created-after, --created-before, --modified-after, and --modified-before (ISO 8601 or YYYY-MM-DD; a bare date covers the whole day and a time without an offset is UTC), and --missing FIELD (repeatable) to select records with no value for a field. Dataset listing adds --name, -m/--measurement, --session, --data-format, --data-type, --instrument-id (resolved to the instrument MFID), --instrument-mfid, --sample-mfid, and -k/--keyword; sample listing adds --name, --type, --description, and --dataset-mfid.
dataset facets FIELD and sample facets FIELD count how many records fall into each value of a grouping field without fetching the records. Dataset fields are session, measurement, data_format, owner, instrument, and project; sample fields are sample_type, owner, and project. Both accept --sort value|label|count, --direction, --limit, --json, and the same project, visibility, ownership, time, exact-match, and --missing filters as the matching list command. In the interactive shell, -m/--measurement, --session, --data-format, and --type complete from these counts, scoped to the --project-id typed on the line or else the current project. These values and project IDs match fuzzily, so xt offers xrd-titan; commands, flags, fixed choices, and MFIDs complete by prefix.
Fields normally updated through dataset update --set include dataset_name, measurement, data_type, session_name, data_format, and timestamp. Use set-public or set-private for public visibility, reassign-project for project changes, and transfer-ownership for owner changes. The CLI has no instrument-assignment command; assign or change a dataset's instrument from Python with client.datasets.assign_instrument().
Sample commands¶
| Command | Description |
|---|---|
sample list |
List samples oldest to newest, in the current project or across every accessible project, with project, type, linked-dataset, description, empty-field, and name-pattern filters |
sample get MFID |
Show a sample and its linked resources |
sample create |
Create a sample |
sample update MFID |
Update sample fields or scientific metadata |
sample edit MFID |
Edit sample fields interactively |
sample reassign-project MFID PROJECT |
Move a sample to another project; PROJECT accepts the project ID or MFID |
sample transfer-ownership MFID USER |
Transfer sample ownership |
sample facets FIELD |
Group samples into value buckets with counts |
sample search QUERY |
Search sample names |
sample search-metadata QUERY |
Search scientific metadata; search-md is an alias |
sample link |
Link parent and child samples; --relationship-type records the kind of link |
sample unlink -p PARENT -c CHILD |
Remove a sample parent-child link |
sample list-parents MFID |
List parent samples; --relationship-type filters by kind of link |
sample list-children MFID |
List child samples; --relationship-type filters by kind of link |
sample link-dataset MFID -d DATASET |
Link a dataset to a sample |
sample unlink-dataset MFID -d DATASET |
Unlink a dataset from a sample |
sample list-datasets MFID |
List datasets linked to a sample |
sample access ... |
List, grant, or revoke direct access entries |
sample set-public MFID |
Make a sample publicly viewable |
sample set-private MFID |
Remove public access from a sample |
Fields normally updated through sample update include sample_name, sample_type, description, and timestamp. Use set-public or set-private for public visibility, reassign-project for project changes, and transfer-ownership for owner changes.
Sample creation accepts --project-id or --project-mfid, including both when they resolve to the same project. Interactive creation continues to prompt for the human-readable project ID.
Project commands¶
| Command | Description |
|---|---|
project list |
List accessible projects; --sort/--direction order them, --member (repeatable) and --member-role narrow to memberships, and a Role column shows your role in each |
project get PROJECT [--include-members] |
Show a project by MFID or project slug, optionally with its members; also shows your role |
project create |
Create a project; --owner USER names the owner (--lead still works with a notice) |
project update ID |
Update a project record or scientific metadata |
project edit ID |
Edit project fields interactively |
project search QUERY |
Search project names and IDs |
project search-metadata QUERY |
Search scientific metadata; search-md is an alias |
project list-users ID |
List project members and roles |
project add-user ID |
Add a user to a project |
project remove-user ID |
Remove a user from a project |
project update-user-role ID USER_ID ROLE |
Change a project member's role |
project transfer-ownership ID USER |
Transfer project ownership |
project request-join ID |
Request membership in a project |
project list-join-requests ID |
List project join requests |
project access ... |
List, grant, or revoke direct access entries |
project set-public ID |
Make a project publicly viewable |
project set-private ID |
Remove public access from a project |
project add-user and project update-user-role require editor or above and accept viewer, contributor, editor, or admin. The target member's current role and requested role must both be below the caller's role. Editors can manage viewers and contributors, admins can also manage editors, and owners can also manage admins. Platform administrators retain their bypass. Use project transfer-ownership to change ownership.
project remove-user permits project owners and platform administrators to remove members, while any member may remove themselves.
Project member tables sort by standing from lead through viewer, display the API owner role as lead, and use semantic role colors when terminal color is enabled: gold for lead, magenta for admin, blue for editor, the default foreground for contributor, and gray for viewer.
The access grant commands accept viewer, contributor, editor, or admin. Use the resource's transfer-ownership command to change ownership. Instruments have no access grant or access revoke; use their member commands.
Instrument commands¶
| Command | Description |
|---|---|
instrument list |
List instruments; --mine lists those you own or are a member of, --sort/--direction order them, --manufacturer, --model, --location, --type filter exactly, --owner (repeatable, up to 10) filters by owner, and a My role column shows your effective role |
instrument get INSTRUMENT |
Show an instrument by MFID or instrument slug (any casing); --include-members adds the members |
instrument create |
Register an instrument |
instrument update MFID |
Update an instrument record or scientific metadata |
instrument set-status MFID STATUS |
Change an instrument lifecycle status |
instrument transfer-ownership MFID USER |
Transfer instrument ownership |
instrument list-service-accounts MFID |
List service accounts bound as instrument operators |
instrument bind-sa MFID SA_MFID |
Bind a service account as an instrument operator |
instrument unbind-sa MFID SA_MFID |
Remove an instrument operator binding |
instrument list-users INSTRUMENT |
List the members, owner first (visible to members) |
instrument add-user INSTRUMENT --user USER [--role viewer\|editor\|admin] |
Add a member (default viewer) |
instrument remove-user INSTRUMENT --user USER |
Remove a member, or yourself |
instrument update-user-role INSTRUMENT USER ROLE |
Change a member's role |
instrument edit MFID |
Edit instrument fields interactively |
instrument search QUERY [--status STATUS] |
Search names, types, and manufacturers with optional lifecycle filtering |
instrument search-metadata QUERY |
Search scientific metadata; search-md is an alias |
instrument access list INSTRUMENT |
List direct access entries |
instrument set-public MFID |
Make an instrument publicly viewable |
instrument set-private MFID |
Remove public access from an instrument |
Instrument commands accept an MFID or the instrument slug, which ignores case. Instrument members work like project members: every member reads the datasets assigned to the instrument, editor also edits its details, and admin also changes status, binds service accounts, and manages members. Admins cannot grant above their own role, and the owner changes only with instrument transfer-ownership. Service accounts are bound with instrument bind-sa rather than added. add-user and update-user-role check the instrument's can_manage_access capability before sending a change.
User commands¶
Most user-management commands require administrator permissions.
| Command | Description |
|---|---|
user get USER |
Show a user by ORCID, MFID, username, or email |
user search QUERY |
Search names and usernames |
user list [--q QUERY] [--first-name NAME] [--last-name NAME] [--username USERNAME] [--email EMAIL] [--platform-role ROLE] [--service-account\\|--human] [--sort FIELD] [--direction DIR] [--exact] [--pending] |
List users; administrators see email and platform roles, and --pending lists accounts waiting for approval |
user create |
Create a user |
user update USER |
Update a user record |
user edit USER |
Edit a user record interactively |
user set-role USER ROLE |
Change a human user's platform role (none, contributor, admin); human platform admins only |
user list-keys USER [--inactive] |
List a user's API keys, never the keys themselves; admin only |
user revoke-keys USER [--key ID] [--yes] |
Revoke every API key held by a user, or one with --key; admin only |
user approve USER |
Let in an account waiting for approval; admin only |
user reject USER [--yes] |
Delete an account waiting for approval, freeing its username and login; admin only |
user list-datasets USER [--limit N\|--all] |
List datasets accessible to a user |
user check-access USER DATASET_MFID |
Show a user's effective dataset access role |
user list-access-groups USER |
List a user's access groups |
user list-projects USER [--role ROLE] [--sort FIELD] [--direction DIR] |
List the projects USER is a member of with USER's role in each, optionally filtered by role (repeatable) and sorted by project_id, title, created, or updated |
Human users require a username and may optionally supply an ORCID during creation. When the ORCID is omitted, the API assigns a canonical MFID. Usernames are normalized to lowercase and must be 3 to 24 characters, start with a letter, contain only letters, digits, underscores, or hyphens, and contain no leading, trailing, or consecutive separators. Interactive creation validates each entry and prompts again when it is invalid.
File commands¶
File commands operate on individual file MFIDs. Dataset-scoped file operations remain available under dataset.
| Command | Description |
|---|---|
file list |
List files globally or within a dataset |
file get ID |
Show file metadata and a download link when available |
file download ID |
Download one file |
file ingestion ID |
Show ingestion requests for a file |
file request-ingestion ID |
Request or repeat ingestion for a cataloged file |
file replace FILE_MFID -i PATH [--ingestor CLASS] [--no-ingest] [--wait] [-y] |
Replace a file's content, keeping its MFID and filename; asks for confirmation unless --yes |
file delete ID [--yes] |
Permanently delete a file after confirmation |
Ingestion commands¶
| Command | Description |
|---|---|
ingestion list |
List ingestion requests |
ingestion get ID |
Show an ingestion request |
ingestion wait ID |
Wait for an ingestion request to finish |
ingestion list-ingestors |
List available ingestion classes |
Service-account commands¶
sa is an alias for service-account. These commands require administrator permissions.
Service-account creation uses the same username rules and interactive validation as human-user creation.
| Command | Description |
|---|---|
sa create |
Create a service account |
sa rotate-key USER |
Generate a new key and revoke every key the account holds |
sa get USER [--groups] |
Show a service account; administrators also see its platform role and API key status, and --groups lists the projects and instruments it belongs to |
sa list [-q TEXT] |
List service accounts with their platform roles; -q filters by username or name |
sa update USER |
Update a service account |
sa edit USER |
Edit a service account interactively |
sa show USER |
Alias for sa get |
sa set-role USER ROLE |
Set a service account's platform role |
sa list-access-groups USER |
List access groups for a service account |
Access-group commands¶
ag is an alias for access-group. Project editors and above review join requests to their projects; platform administrators review all of them.
| Command | Description |
|---|---|
ag request PROJECT |
Request to join a project |
ag mine |
List the current user's join requests |
ag list |
List join requests you may review; --scope, --project, --status (repeatable), --search, --sort |
ag get ID |
Show a join request |
ag approve ID... |
Approve pending join requests; --role sets the granted role (default contributor) |
ag reject ID... |
Reject pending join requests |
ag withdraw ID... |
Withdraw your own pending join requests |
Account commands¶
These commands operate on the currently authenticated account and do not require administrator permissions.
| Command | Description |
|---|---|
login [--device\|--browser\|--paste] [--no-open] |
Sign in with a browser; --device approves a code on any device, --paste reads the code the sign-in page shows |
logout |
End and delete the stored sign-in for the configured API |
account show |
Show the current profile |
account update |
Update profile fields |
account edit |
Edit the current profile interactively |
account keys list [--inactive] |
List your API keys, never the keys themselves |
account keys create NAME [--expires DATE] |
Create an API key and show it once; needs a recent browser sign-in, not an API key |
account keys rename ID NAME |
Rename one of your API keys |
account keys revoke ID [--yes] |
Revoke one of your API keys |
account verify |
Check API-key validity and expiry |
Treat output from account keys create as a secret. Do not include it in logs, issue reports, or agent prompts.
Deletion commands¶
The deletion-request workflow is separate from direct permanent deletion. Review and audit commands require administrator permissions.
| Command | Description |
|---|---|
deletion request RESOURCE |
Request deletion of a dataset or sample |
deletion list |
List deletion requests |
deletion get ID |
Show a deletion request |
deletion approve ID... |
Approve pending requests |
deletion reject ID... |
Reject pending requests |
deletion delete RESOURCE |
Permanently delete a resource |
deletion list-deleted |
List permanent-deletion audit records |
deletion get-deleted ID |
Show a permanent-deletion audit record |
deletion list accepts --scope accessible|submitted|reviewable, --project-mfid, --sort request_time|resource_name|project_id|requester_name|status, and --direction. deletion list-deleted accepts --scope all|submitted, --project-id, --project-mfid, and --direction. deletion get shows whether you may review the request.
Cast command¶
cast loads a declarative .crux recipe containing datasets, samples, files, and links.
| Command | Description |
|---|---|
cast FILE |
Execute a recipe |
cast FILE --validate |
Validate references and cycles without executing |
cast FILE --dry-run |
Preview execution without API mutations |
cast FILE --show |
Show the plan and lock status |
cast FILE --force |
Clear the lock and recreate entities |
cast FILE --reupload |
Re-upload files without recreating records |
Configuration and cache¶
| Command | Description |
|---|---|
config init |
Run interactive configuration setup |
config show |
Show the current configuration |
config get KEY |
Print one configuration value |
config set KEY VALUE |
Set one configuration value |
config unset KEY |
Remove a config-file override and return to the environment or package default |
config path |
Show the configuration-file path |
config edit |
Edit the configuration file |
cache show |
Show cache location and disk usage |
cache clear |
Remove cached files by dataset, age, or all entries |
Configuration values can come from environment variables, the platform-specific config file, or defaults. Avoid displaying api_key in shared terminals or logs.
dataset delete, file delete, and cache clear prompt before removing data. Use --yes only when the operation has already been explicitly approved, such as in a controlled noninteractive workflow.
General utility commands¶
| Command | Description |
|---|---|
status |
Show endpoint reachability, deployment provenance, database readiness, and authentication identity |
whoami |
Show the identity, platform role, privilege mode, and capabilities of the active credential, and whether it is an API key or a sign-in and where it came from |
get MFID |
Show a dataset, sample, project, or instrument after detecting its resource type |
edit MFID |
Edit a dataset, sample, or instrument after detecting its type |
download MFID |
Save a record and, for datasets, associated files |
link |
Link parent-child resources or associate a dataset and sample; --relationship-type applies to parent-child links only |
unlink MFID1 MFID2 |
Remove a resource relationship |
tree MFID |
Display connected ancestors and descendants |
open [ID] |
Open the Crucible Hub, or a resource's Hub page, or print its URL |
qr ID |
Print a terminal QR code for an MFID |
print barcode PRINTER_ID MFID NAME |
Publish a barcode print job and wait for confirmation it printed; exits non-zero unless the printer confirms success |
completion [SHELL] |
Generate and install completion for bash, zsh, fish, or tcsh |
Renamed in 4.0.0¶
The old names still work and print a warning naming the new one; they are left out of help and completion.
| Old | New |
|---|---|
dataset add-sample, dataset remove-sample |
dataset link-sample, dataset unlink-sample |
dataset remove-child PARENT -c CHILD |
dataset unlink -p PARENT -c CHILD |
sample add-dataset, sample remove-dataset |
sample link-dataset, sample unlink-dataset |
sample remove-child PARENT -c CHILD |
sample unlink -p PARENT -c CHILD |
dataset ingestors |
ingestion list-ingestors |
Removed in 4.0.0¶
These commands and options stop with the form to use instead.
| Removed | Use instead |
|---|---|
upload ... |
dataset create -i FILE ... |
account api-key |
account keys create NAME |
dataset list-access-groups, dataset add-access-group |
dataset access list, dataset access grant |
user add-access-group, user remove-access-group |
project add-user/remove-user, or instrument bind-sa/unbind-sa |
sa add-access-group, sa remove-access-group |
instrument bind-sa/unbind-sa, or project add-user/remove-user |
RESOURCE publish, RESOURCE unpublish |
RESOURCE set-public, RESOURCE set-private |
dataset update-metadata, dataset get-keywords |
dataset update --metadata, dataset list-keywords |
user get-access-groups, user get-projects, project get-users |
user list-access-groups, user list-projects, project list-users |
--orcid, --username, --email on user get/edit |
the USER argument |
--unique-id, --username on sa commands |
the SA argument |
--orcid, --email, --username on project add-user/remove-user |
--user |
-pid, -id, --project (as project ID) |
--project-id / -p |
--instrument NAME |
--instrument-id or --instrument-mfid |
project get --members |
--include-members |
instrument get --by-id |
drop it; MFIDs and instrument IDs are told apart automatically |
sample update --public/--no-public |
sample set-public/set-private |
--upload-protocol |
drop it; every upload uses signed multipart |
ag list --group |
--project |
CRUCIBLE_CURRENT_PROJECT |
--project-id, or use PROJECT_ID in the shell |
instrument access grant, instrument access revoke |
instrument add-user, update-user-role, remove-user |
dataset update --set public=... |
dataset set-public, dataset set-private |