Generated from hq-cli src/command-catalog.generated.ts at hq-cli 5.345.177 by scripts/generate-cli-reference.mjs. Do not edit this page by hand; re-run the script instead.
Hidden and staff-only commands are not listed. Run hq <command> --help for the full help text of any command, including options of nested subcommands.
Manage file access controls in HQ vault. A trailing folder slash (foo/) is a private folder; the quoted glob ‘foo/’ shares all children, and a quoted ‘clients//reports’ shares that folder under every client.
Tell whether a vault file exists and whether you can read it. Never a bare not-found. Takes one or more paths or queries. Exit codes: 0 local/not-synced, 2 never-existed, 3 no-access, 4 ambiguous.
Send and read direct messages, and manage connection requests. Messages sent from a person login are always tagged for people. When an agent or bot sends, messages to an agent, a channel or a group are tagged for agents and folded away from people by default. Pass —for-human or —audience human to tag a message for people; —audience agent|both is for agents and bots only.
With —global Claude: install directly or via plugin (direct | plugin) Default: "direct".
--marketplace-dir <dir>
With —via plugin: local Claude Code marketplace directory (or HQ_CLAUDE_PLUGIN_MARKETPLACE)
--dry-run
With —global: print every file and config diff without writing
--hq-root <dir>
With —global: HQ checkout to install from (default: walk up from cwd)
hq remove
Remove an installed package (archives it first)
hq remove [options] <slug>
hq publish
Package a skill/worker pack and submit it to the HQ marketplace (POST /v1/listings).
hq publish [options] <path>
Option
Description
--as <handle>
Publish on behalf of a seeded, unclaimed creator by handle (HQ staff only). The listing is attributed to that creator instead of you.
--on-behalf <handle>
Alias for —as.
--handle <handle>
Publish as one of your own claimed creator handles (for accounts that own more than one). Defaults to your first claimed handle.
hq creators
Creator marketplace account
hq creators [options] [command]
Subcommand
Description
apply [options]
Apply for verified-creator access (required to publish packs)
hq sync
Cloud sync commands — sync HQ to S3 for mobile access
hq sync [options] [command]
Subcommand
Description
push [options] [paths...]
Push local file(s) to the company vault on S3
pull [options]
Pull permitted files from the company vault to local HQ
status [options]
Show local sync journal summary
now [options]
Bidirectional sync: push local changes, then pull remote updates (mirrors AppBar HQ Sync’s “Sync Now” button)
doctor [options]
Repair the local HQ tree. --reconcile-conflicts folds legacy sibling .conflict-* twins back into their live files (higher frontmatter version: wins; losers are parked under .hq/conflict-backups/). Dry-run unless —yes.
mode [options] [mode]
Flip a membership’s sync-mode (shared|all|custom), or —show every membership’s current mode
narrow [options]
Migrate a company’s membership from syncMode=‘all’ to ‘shared’ — dry-run previews diff, —apply performs the prune + flip
manifest [options]
Upload a file manifest for the sync reconciliation audit (personal or a company scope)
held [options]
Show uploads held because someone else was signed in on this computer
release-held [options]
Confirm held files are yours and let sync upload them
hq sync push
Push local file(s) to the company vault on S3
Option
Description
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
--company <slug>
Company slug or UID (defaults to active company in .hq/config.json)
--lock-timeout <seconds>
Maximum wait for another active HQ operation (default: 300; 0 = refuse immediately)
--message <msg>
Optional message attached to journal entries for these uploads
--on-conflict <strategy>
Conflict strategy: overwrite | keep | abort (omit for interactive)
--creds-from-stdin
Read a pre-vended EntityContext as JSON from stdin instead of vending via the cached Cognito session. Use when the caller (e.g. AppBar HQ Sync) has its own STS pipeline (/sts/vend-child with task scope) and just needs share()‘s upload mechanics. The caller is responsible for vending credentials with enough TTL for the run.
--json
Emit each share()-level event as a JSON Lines record on stderr (one JSON object per line) instead of human-readable console output. A synthetic {type:"complete",...} line is appended at the end with the final ShareResult. Subprocess callers parse these to render their own UI (e.g. AppBar Tauri events).
--all
Push every company you are a member of plus your personal vault. Company targets push <hq-root>/companies/<slug>/; personal pushes every top-level entry under <hq-root> minus the excluded set (.git, companies, repos, workspace). Implies —skip-unchanged + —propagate-deletes. Mutually exclusive with —company, —personal, and any explicit [paths].
--personal
Push to the caller’s canonical personal vault (resolved from the cached Cognito session). When no [paths] are given, defaults to every top-level entry under <hq-root> minus the excluded set (.git, companies, repos, workspace) — same scope as --all’s personal slot. Mutually exclusive with —company and —all.
--no-personal
In --all mode, skip the canonical-person leg of the fanout — only push to the caller’s company memberships. Mirrors the upstream hq-sync-runner --skip-personal flag (hq-cloud 5.25.0); wired so the AppBar HQ Sync menubar can drop personal sync via a toggle, and so CLI users can opt out one-off. Ignored outside --all.
hq sync pull
Pull permitted files from the company vault to local HQ
Option
Description
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
--company <slug>
Company slug or UID (defaults to active company in .hq/config.json)
--lock-timeout <seconds>
Maximum wait for another active HQ operation (default: 300; 0 = refuse immediately)
--on-conflict <strategy>
Conflict strategy: overwrite | keep | abort (omit for interactive)
--all
Pull every company you are a member of plus your personal vault into <hq-root>. Companies land at <hq-root>/companies/<slug>; the personal vault syncs at <hq-root>. Mutually exclusive with —company and —personal.
--personal
Pull the caller’s canonical personal vault into <hq-root> directly (no companies/<slug>/ prefix). Resolves the person UID automatically from the cached Cognito session. Mutually exclusive with —company and —all.
--no-personal
In --all mode, skip the canonical-person leg of the fanout — only pull the caller’s company memberships. Mirrors the upstream hq-sync-runner --skip-personal flag (hq-cloud 5.25.0). Ignored outside --all.
--mode-all
US-011: opt out of the strict narrow-hint refusal for this run. Has no effect today (default narrow-hint level is ‘hint’); wired so future hq-core-staging releases can flip the default to ‘strict’ without re-touching this command.
--force-scope-shrink
When a scope shrink would un-track locally-modified files (e.g. after narrowing from all → shared), proceed anyway. Dirty files are KEPT on disk and only un-tracked from sync; clean out-of-scope files are quarantined under .hq/scope-quarantine/ (recoverable).
hq sync status
Show local sync journal summary
Option
Description
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
--company <slug>
Show only this company’s journal (company slug)
hq sync now
Bidirectional sync: push local changes, then pull remote updates (mirrors AppBar HQ Sync’s “Sync Now” button)
Option
Description
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
--company <slug>
Company slug or UID (defaults to active company in .hq/config.json)
--lock-timeout <seconds>
Maximum wait for another active HQ operation (default: 300; 0 = refuse immediately)
--message <msg>
Optional message attached to journal entries for the push leg
--on-conflict <strategy>
Conflict strategy: overwrite | keep | abort (omit for interactive)
--all
Sync every company you are a member of plus your personal vault (pushAll then pullAll). Mutually exclusive with —company and —personal.
--personal
Sync the caller’s canonical personal vault bidirectionally. Mutually exclusive with —company and —all.
--no-personal
In --all mode, skip the canonical-person leg of both the push and pull fanouts — only sync the caller’s company memberships. Mirrors the upstream hq-sync-runner --skip-personal flag (hq-cloud 5.25.0). Ignored outside --all.
--mode-all
US-011: opt out of the strict narrow-hint refusal for this run. No-op today; wired so future hq-core-staging releases can flip the default narrow-hint level to ‘strict’.
--force-scope-shrink
When a scope shrink would un-track locally-modified files, proceed anyway. Dirty files are KEPT on disk (only un-tracked); clean out-of-scope files are quarantined under .hq/scope-quarantine/.
--max-idle <seconds>
For scheduled or unattended runs: stop with exit code 124 after this many seconds with no file planned, transferred or skipped. The remote listing and planning phase of a leg reports no progress, so set this longer than your slowest planning phase. Laptop sleep does not count. (default: HQ_SYNC_MAX_IDLE, else off)
--max-runtime <seconds>
For scheduled or unattended runs: stop with exit code 124 if the run has not finished after this many seconds, even while it is making progress. Laptop sleep does not count. (default: HQ_SYNC_MAX_RUNTIME, else off)
hq sync doctor
Repair the local HQ tree. --reconcile-conflicts folds legacy sibling .conflict-* twins back into their live files (higher frontmatter version: wins; losers are parked under .hq/conflict-backups/). Dry-run unless —yes.
Option
Description
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
--reconcile-conflicts
Reconcile legacy sibling .conflict-<ts>-<machine> twins against their live files (purely local; no vault access).
--yes
Apply the plan (default: dry-run, print only)
hq sync mode
Flip a membership’s sync-mode (shared|all|custom), or —show every membership’s current mode
Option
Description
--company <slug>
Company slug (defaults to the active company in <hq-root>/.hq/config.json)
--show
Print the current sync-mode for every membership the caller has
--paths <csv>
Comma-separated allowed prefixes (required when mode is ‘custom’)
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
hq sync narrow
Migrate a company’s membership from syncMode=‘all’ to ‘shared’ — dry-run previews diff, —apply performs the prune + flip
Option
Description
--dry-run
Compute + print the narrow plan; write nothing (default if neither —dry-run nor —apply is passed)
--apply
Execute the narrow — delete clean orphans, flip sync-mode to ‘shared’
--yes
Skip the interactive confirmation prompt on —apply
--force
On —apply, also remove locally-modified (dirty) files (destructive — no undo)
--company <slug>
Company slug (defaults to the active company in <hq-root>/.hq/config.json)
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
hq sync manifest
Upload a file manifest for the sync reconciliation audit (personal or a company scope)
Option
Description
--scope <scope>
’personal’ or a company slug (defaults to the active company, else personal)
--full
Send a fresh full baseline instead of a delta
--print
Build and print the chunks as JSON; upload nothing
--json
Print the pass outcome as one JSON line instead of a summary
--respect-throttle
Honour the 24h per-scope upload throttle instead of forcing a pass (for scheduled callers)
--engine-owned
The new sync engine owns this scope here: build the manifest from disk alone and mark the legacy journal as not its ledger
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
hq sync held
Show uploads held because someone else was signed in on this computer
Option
Description
--json
Print the held scopes as JSON
hq sync release-held
Confirm held files are yours and let sync upload them
Option
Description
--company <slug>
Release only this company
--personal
Release only the personal vault
--yes
Do not ask for confirmation
--json
Print the result as JSON
hq cloud
Cloud commands — provision entities and manage cloud-backed companies
hq cloud [options] [command]
Subcommand
Description
provision [options] [command]
Provision a cloud-backed entity (entity + bucket + initial sync)
provision company [options] <slug>
Promote a local company to a cloud-backed entity (idempotent). Provisions the vault entity if missing, patches manifest.yaml, writes .hq/config.json, and triggers an initial sync.
retire [options] [command]
Soft-tombstone a cloud-backed company (owner only)
retire company [options] <slug>
Soft-tombstone a cloud company so hq cloud demote company can run. Does not delete vault files or local folders. Prompts unless —yes.
demote [options] [command]
Demote a cloud-backed entity back to local-only
demote company [options] <slug>
Demote a cloud-backed company to local-only after the cloud entity has been soft-tombstoned (hq cloud retire company or hq-console). Removes .hq/config.json, flips company.yaml cloud: false, and strips the manifest cloud refs.
hq team-sync
Pull latest team content for all joined teams
hq team-sync [options]
Option
Description
--team <slug>
Sync only a specific team by slug
--dry-run
Show what would be synced without making changes
hq login
Authenticate with HQ via Cognito
hq login [options]
Option
Description
--provider <provider>
OAuth provider to use: google, microsoft, or picker
--no-browser
Print the sign-in URL without opening a browser
hq logout
Log out and clear cached credentials
hq logout [options]
hq whoami
Show the currently authenticated user
hq whoami [options]
Option
Description
--json
Output identity metadata as JSON (never tokens)
--company <slug>
Also report this company’s billing and usage notice
hq link
Link this folder (or its git remote) to an HQ company
hq link [options] [company]
Option
Description
--path <dir>
Folder to link (default: current directory)
--list
Print every linked folder
--json
Print JSON output
hq unlink
Remove the HQ company link for this folder
hq unlink [options]
Option
Description
--path <dir>
Folder to unlink (default: current directory)
hq uninstall
Remove the HQ global install from an agent runtime (use with —global)
hq uninstall [options]
Option
Description
--global
Reverse hq install --global
--runtime <runtime>
Runtime to uninstall from (claude | codex) Default: "claude".
--dry-run
Print what would change without writing
--hq-root <dir>
HQ checkout (only needed when install state is missing)
hq resolve-company
Print the HQ company linked to this folder (never guesses)
hq resolve-company [options]
Option
Description
--path <dir>
Folder to resolve (default: current directory)
--json
Print JSON output
hq provider
Launch supported AI provider CLIs
hq provider [options] [command]
Subcommand
Description
launch [options] <name> [args...]
Run a noninteractive Claude Code or Codex command with sign-in guidance
hq version
Show HQ, CLI, and latest hq-core versions
hq version [options]
Option
Description
--json
Output the same payload as JSON
hq fingerprint
Print this session’s HQ fingerprint, or resolve a fingerprint to its session
hq fingerprint [options] [command] [value]
Option
Description
--json
output JSON
Subcommand
Description
self [options]
Print the fingerprint for the current HQ session
hq capture
Record your screen with HQ Capture: start, record, stop, status and wait
hq capture [options] [command]
Subcommand
Description
doctor [args...]
Check this Mac can build and run HQ Capture
build [args...]
Build HQ Capture if missing or out of date ([—force])
start [args...]
Open HQ Capture for the session’s company ([—company <slug>] [—record])
record [args...]
Start recording in the open HQ Capture, as if Start was pressed
status [args...]
Print HQ Capture’s current status as JSON
wait [args...]
Wait until the recording is saved, failed or closed ([—timeout <seconds>])
stop [args...]
Stop HQ Capture, saving a recording in progress
run [args...]
Run an agent command on a saved recording (<mode> <package> [args])
resume [args...]
Check a pasted resume prompt and print its package (<prompt>)
hq auth
Manage the local HQ Cognito session
hq auth [options] [command]
Subcommand
Description
machine-session [options]
Write the existing machine identity session to a private file, or to a pipe with —output -
login [options]
Sign in to HQ — opens the Cognito Hosted UI and caches tokens locally
logout [options]
Clear the cached HQ Cognito session
keychain [options]
Show or retry HQ’s access to your sign-in in the macOS Keychain
refresh [options]
Ensure a valid cached Cognito session (refresh only if expiring; no-op if already valid)
status [options]
Show whether a valid HQ session is cached
hq secrets
Manage secrets in HQ vault (SSM Parameter Store)
hq secrets [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
--personal
Operate on the caller’s personal vault (no sharing)
Subcommand
Description
set [options] <name>
Create or update a secret
get [options] <name>
Get a secret’s metadata (use —reveal for value)
exists [options] <name>
Check whether a secret exists (HEAD; exit 0=present, 1=absent, 2=error)
list [options]
List all secrets for the company (including nested path-based names)
policy [options] [command]
View or update secret access policy
policy get [options] <path>
Show the policy for a secret path
policy set [options] <path>
Set the policy for a secret path
script [options] [command]
Manage approved scripts for script-locked secrets
script approve [options] <path>
Approve a script for a secret path. Each —id holds one approval: approving an id that is already approved for a different path or hash fails unless —replace is given. To let the same script run from several machines, approve one HQ-relative —remote-path instead of one absolute path per machine.
script revoke [options] <path>
Revoke an approved script from a secret path
script list [options] <path>
List approved scripts for a secret path
delete [options] <name>
Delete a secret
sandbox [options]
Run a command in the hosted sandbox with named secrets injected as env vars; open egress, secrets never touch this machine
Generate a one-time link for someone to submit a secret value
share [options] <path>
Share a secret with an email address, group, fleet agent (agt_…), company Outpost (out_…), or the entire company (@all)
unshare [options] <path>
Remove a grant from a secret
acl [options] <path>
Show the ACL (access control list) for a secret path
cache [options] [command]
Manage the local secrets cache
cache clear [options]
Clear all cached secrets
hq db
Vault databases — local SQLite per company and HQ-managed remote Postgres-class DBs
hq db [options] [command]
Subcommand
Description
status [options]
Show company local/remote status or your personal database status
sql [options] [sql...]
Run read-only SQL against a company or personal database; use —write for changes
migrate [options]
Apply pending vault text migrations (companies/{co}/db/migrations/*.sql) to the local DB
provision [options]
Create a company remote database or your personal database
usage [options]
Show this month’s remote database usage. With —app: included, used, overage, ceiling and the estimated charge for one app
dump [options]
Export an app’s remote database schema and data to a SQL file
destroy [options]
Permanently delete an app’s remote database (owners and admins only; cannot be undone)
hq api-keys
Manage vault API keys for CI and automation
hq api-keys [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
create [options]
Create a new API key (vault secrets and/or scoped deploy via —deploy-app)
list [options]
List API keys for a company
revoke [options] <keyId>
Revoke an API key
hq run
Load secrets from .env.schema and run a command with them injected
hq run [options]
Option
Description
--company <slug>
Company slug (overrides @hqCompany in schema)
--schema <path>
Explicit schema path (skips walk-up discovery)
--script <path>
Attach local script identity for script-locked secrets
--script-id <id>
Stable script identifier approved by policy
--check
Resolve schema and validate vars without executing the command
hq groups
Manage groups in HQ vault
hq groups [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
create [options] <groupId>
Create a new group
delete [options] <groupId>
Delete a group
add [options] <groupId> <principal>
Add a person, agent, or company Outpost to a group (principal: email, personUid, agentUid, or out_ Outpost id)
remove [options] <groupId> <principal>
Remove a person, agent, or company Outpost from a group (principal: email, personUid, agentUid, or out_ Outpost id)
list [options]
List all groups in the company
members [options] <groupId>
List members of a group
hq workers
Discover and share HQ workers
hq workers [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
list [options]
List workers you can access (public + your active company’s)
share [options] <workerId>
Grant a teammate, group, or @all access to a company worker
hq group-grants
Grant a group’s access to another company (cross-company), and revoke or inspect those grants
hq group-grants [options] [command]
Option
Description
--company <slug>
Source company slug (the company that owns the group; resolves to sourceCompanyUid)
Subcommand
Description
grant [options] <groupId> <targetCompany>
Grant a group (from the source company) access to <targetCompany> at a role
revoke [options] <groupId> <targetCompany>
Revoke a group’s grant on <targetCompany>
outbound [options]
List grants a source-company group holds on other companies; requires —group <groupId>.
inbound [options]
List grants other companies’ groups hold on this company
hq files
Manage file access controls in HQ vault. A trailing folder slash (foo/) is a private folder; the quoted glob ‘foo/’ shares all children, and a quoted ‘clients//reports’ shares that folder under every client.
hq files [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
share [options] [paths...]
Share file paths. foo/ is a private folder (children stay creator-only); quote the shared-folder glob ‘foo/’, or a one-folder wildcard such as ‘clients//reports’, so the shell passes it literally. Without —with: mint a share-session URL and open it in the browser. With —with: grant access directly to a person, group, or @all.
unshare [options] <prefix>
Remove a named file access grant, or remove an empty ACL row or grantee-less entries without —with. With —with, a bare name such as ‘knowledge’ first removes the exact ‘knowledge’ grant (rows from older hq files share versions) and, when none matches, the folder-glob ‘knowledge/’ grant. Without —with, the prefix is used exactly as typed: pass ‘knowledge/’ (quoted) to clean up the folder-glob row.
acl [options] <prefix>
Show the ACL (access control list) for a file prefix. foo/ is a private folder whose children are creator-only; the quoted glob ‘foo/*’ is shared.
delete [options] <prefix>
Delete vault objects by exact key or bounded prefix. A trailing slash expands recursively; use * for an explicit wildcard prefix. Always previews the exact count first; prompts for confirmation unless —yes. Use —personal to target your own personal vault instead of a company.
versions [options] <path>
List prior content versions and delete markers for one exact vault key. Restore a listed version with restore —version-id <id>. Use —personal for your personal vault.
restore [options] <path>
Restore a prior version with —version-id or undelete an exact vault key. Prompts before overwriting unless —yes. Use —personal for your personal vault.
trash [options]
List deleted vault keys retained as tombstones. Use —personal for your personal vault.
browse [options] [path]
List vault objects under [path] without syncing them locally. Private-folder rows (foo/) are labelled and do not imply child access; their children are creator-only. Uses the browse-vend path (role-bypass allowed). Pass —personal to browse the caller’s personal vault; otherwise [path] must start with companies/<slug>/.
cat [options] <path>
Stream a single vault object to stdout (or —out <file>) without syncing it. Uses the browse-vend path. Pass —personal to read from the caller’s personal vault.
shared-with-me [options]
List the files/prefixes explicitly shared with you. Private folders (foo/) expose only the folder; children are creator-only. Omit —company to roll up across every company you’re a member of. Pure read — no download, no credentials vended. Owner/admin role-bypass access is NOT listed (only explicit grants).
search [options] <query>
Search vault object keys (case-insensitive path/name match) under a company without downloading. Requires —company (or —personal). Content search is not supported in v1.
get [options] <path>
Download (materialize) a vault file or prefix into local HQ on demand. Default writes in place under <hqRoot>/companies/<slug>/<path> and registers a pin so a scoped sync (mode shared|custom) won’t prune it. Use —into to write elsewhere (no pin). Company mode only.
hq access
Tell whether a vault file exists and whether you can read it. Never a bare not-found. Takes one or more paths or queries. Exit codes: 0 local/not-synced, 2 never-existed, 3 no-access, 4 ambiguous.
hq access [options] [vault-path...]
Option
Description
--company <slug>
Company slug
--as <person>
Check a person’s access (email, prs_, or agt_ uid; self or owner/admin)
--json
Print a single JSON object (an array for several paths)
--stdin
Read more paths from stdin, one per line
--no-fix
Diagnose only: do not fetch, pin, or repair sync
--yes
Answer the one send-confirmation yes (hosts must pass this explicitly)
--hq-root <path>
Local HQ tree root (default: <runtime-path>)
hq storage
See how much space HQ backups use and delete old ones
hq storage [options] [command]
Subcommand
Description
status [options]
Show local autosave history and cloud file history sizes
big-files [options]
Show big files that could move to HQ cloud (the slow scan status skips; read-only)
prune [options]
Delete local autosave history and old cloud file versions before a date
offload [options]
Move big files to HQ cloud storage and leave links behind (old versions in history; with —current, idle working files to the vault that owns them)
fetch [options] [path]
Download a file moved to HQ cloud storage, check it, and put it back in place
hq skill
Create, stage, promote, and discuss company skills
hq skill [options] [command]
Option
Description
--company <slug>
Company slug (defaults to the active company)
--hq-root <path>
Local HQ root Default: "<runtime-path>".
Subcommand
Description
create [options] <slug>
Register, stamp, surface, and sync a company skill
register [options] <slug>
Register a company skill to a proposal-lane stamped file without writing the canonical tree
promote [options] <slug>
Copy independently-cleared stamped bytes into the canonical skill path after SHA-256 and skill_uid checks
propose [options] <target>
Post a comment-only improvement for a skill
delete [options] <target>
Delete a company skill and all of its files for everyone (skl_… uid, SKILL.md path, or skill slug). Prompts for confirmation unless —yes.
hq members
Manage company memberships and invites
hq members [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
promote [options] <target> <newRole>
Change a member’s role (owner|admin|member|guest) — promotes OR demotes. <target> may be an email, a prs_ personUid, or a full membership key. Owner-or-admin only; setting a target to owner (or changing an owner) is owner-only. Server-enforced.
invite [options] <target>
Invite a person or fleet agent to the company. People: email or prs_…. Agents: agt_… or agt-…@agents.getindigo.ai (member|admin; host keeps billing, no double charge).
list [options]
List the company’s active members (use —pending for pending invites)
remove [options] <target>
Remove an active member. Ends membership so company-wide grants no longer apply, and drops that person’s file and secret grants. Accepts an email, prs_ personUid, agt_ agentUid, or full membership key.
revoke [options] <target>
Revoke a pending invite. For an active member, use hq members remove instead. Accepts an email, personUid, full membershipKey, or legacy inviteToken.
hq people
List, search, and resolve a company’s people (from companies/<co>/people)
hq people [options] [command]
Option
Description
--company <slug>
Company slug to scope to (defaults to the active company)
Subcommand
Description
list [options]
List all people recorded for the company
search [options] <keyword>
Keyword search over people names and emails
resolve [options] <name>
Resolve a person name to their email address
hq dm
Send and read direct messages, and manage connection requests. Messages sent from a person login are always tagged for people. When an agent or bot sends, messages to an agent, a channel or a group are tagged for agents and folded away from people by default. Pass —for-human or —audience human to tag a message for people; —audience agent|both is for agents and bots only.
hq dm [options] [command]
Subcommand
Description
forward [options] <eventId>
Forward a DM or channel message through the existing send routes.
inbox [options]
List your recent incoming direct messages.
thread [options] <person>
Show your two-way conversation with a person (email, personUid, or agentUid). Reads oldest-first and marks their messages read unless —no-ack.
channel [options] <target>
Show recent messages in a DM channel or group DM — by name, #name, or a channel id from hq channels. Roots print their eventId; replies indent under their root. —json is the raw oldest-first array (ungrouped).
requests [options]
List your pending incoming connection requests.
accept [options] <person>
Accept a pending connection request (by requester email, personUid, or pairKey from hq dm requests). Their held message is then delivered.
decline [options] <person>
Decline a pending connection request (by requester email, personUid, or pairKey from hq dm requests).
block [options] <person>
Block a person so they can’t send you further requests (by requester email, personUid, or pairKey from hq dm requests).
hq channels
List the DM channels you’re in (name them with hq dm <name> "…").
hq channels [options] [command]
Subcommand
Description
list [options]
List your DM channels — named channels and group DMs.
changes [options]
Print the channels that changed since a feed cursor, as JSON (no cursor → full snapshot).
members [options] <channel>
Print the members of one of your DM channels.
delete [options] <channel>
Delete a channel you own, including its messages. Group DMs cannot be deleted.
hq onboard
Provision an HQ vault: sign in, create company, S3 bucket, STS, sync
hq onboard [options] [command]
Subcommand
Description
create-company [options]
Sign in and provision a brand NEW HQ vault for a company. To join a teammate’s EXISTING company instead, use hq onboard join — creating a same-named company makes a separate one you’d be alone in.
join [options]
Accept an invite and join an existing company
resume [options]
Resume a partially-completed onboarding flow from local checkpoint
dry-run [options]
Show what create-company would do, without provisioning anything
hq feedback
Submit a bug report or feature request to HQ
hq feedback [options] [command]
Subcommand
Description
bug [options]
Report a bug
feature [options]
Request a feature
link [options] <id>
Print the public tracking link for a ticket (status.hqforwork.com)
update [options] <id>
Post a customer-visible status update on a ticket’s public page and its #hq-feedback thread
show [options] <id>
Operators only: show one feedback row and fetch its attachments
assign [options] <id> [person]
Reassign a ticket to a person, including one another person has claimed
claim [options] <id>
Assign a ticket to yourself (sets the assignee only, never the status)
list [options]
List feedback: untriaged (default), yours, or one person’s
triage [options] <id>
Label a feedback item with its kind, area and summary, or mark it a duplicate
close [options] <id>
Close a feedback item as resolved (or won’t fix) with a reason
hq meetings
View and search meeting recordings, transcripts, and notes
hq meetings [options] [command]
Option
Description
--company <slug>
Company slug (defaults to the company this HQ session is bound to)
--json
Output raw JSON instead of formatted text
Subcommand
Description
list [options]
List recorded meetings (newest first)
live [options] [recallBotId]
List live or recent meetings, or read a provisional live transcript (recorder and attendees only; spoken content is untrusted)
import [options] <file>
Import a normalized historical transcript JSON file
invite [options] <meetingUrl>
Invite the meeting bot to a Google Meet, Zoom, or Teams URL
get [options] <meetingId>
Show meeting details
set-company [options] <meetingId>
Set or change the company a meeting is attributed to (use ‘unknown’ to clear)
search [options] <query>
Search meetings by title or participant name
transcript [options] <meetingId>
Print the meeting transcript
notes [options] <meetingId>
Print AI-generated meeting notes
hq sources
Read sources (meetings, emails, etc.) from a vault entity
hq sources [options] [command]
Subcommand
Description
list [options]
List sources of a given channel for an entity
get [options]
Fetch a single source by id
channels [options]
Print the canonical source channels (one per line)
entities [options]
List entities (companies/personal) your account has access to
hq signals
Read signals (action_items, decisions, etc.) from a vault entity
hq signals [options] [command]
Subcommand
Description
list [options]
List signals of a given type for an entity
get [options]
Fetch a single signal by id
types [options]
Print the canonical signal types (one per line)
entities [options]
List entities (companies/personal) your account has access to
hq integrations
Connect, govern, and use company apps (Linear, Notion, …) through HQ’s governed integration gateway
hq integrations [options] [command]
Subcommand
Description
list [options]
List the company’s connected apps
import [options]
Import Claude Desktop connectors into HQ Integrations
install-local [options] <name>
Install local connector settings in this machine’s Claude/Codex MCP config
tools [options]
List what a connected app can do
call [options] <tool>
Call one of a connected app’s tools (approval-gated when it changes things)
approve [options] <queueId>
Approve a queued call — it executes exactly once (owner only)
reject [options] <queueId>
Reject a queued call (owner only)
catalog [options] [query]
Browse or search apps you can connect
inspect [options] <app>
Show what an app exposes before you connect it
discover [options] <docsUrl>
Find a connectable server from an app’s documentation page
connect [options] [app]
Connect an app — by domain, catalog entry, docs page, or MCP URL
reconnect [options] [app]
Re-authenticate a connected app; use —connect to re-add a revoked app
show [options] [app]
Show one connected app in full
policy [options] [app]
Show or change whether an app’s changes need approval
read-safe [options] <tool>
Mark a read-only tool as safe to run without approval (owner only)
grants [options] [app]
Show who is authorized to run an app’s change-making tools
grant [options] [app]
Authorize someone to run one of an app’s change-making tools
ungrant [options] [app]
Remove a per-tool authorization, leaving only the app’s default access
access [options] [app]
Show who in the company can use a connected app
share [options] [app]
Let someone in the company use a connected app
unshare [options] [app]
Stop someone from using a connected app
audit [options]
Recent activity across the company’s connected apps
pending [options]
Calls that were queued for an owner’s approval
disconnect [options] [app]
Disconnect an app and delete its stored credentials
purge [options] [app]
Permanently remove a revoked connection from the list (owner only)
slack [options] [command]
Connect your own Slack account to HQ (stored in your personal vault)
slack connect [options]
Get Slack’s sign-in link for your account; the sign-in finishes on HQ
slack accounts [options]
List your connected Slack accounts
hq import
Import existing AI context (Claude Code, Codex, repos) into HQ
hq import [options] [command]
Subcommand
Description
scan [options]
Scan this machine for AI context (sessions, repos, companies, projects); local only
hq scan
Local, offline scanners
hq scan [options] [command]
Subcommand
Description
secrets [options] [files...]
Scan staged changes (or files) for secrets; exits 1 on a finding
hq reindex
Surface namespaced skills, materialize legacy knowledge repos, regenerate the workers registry, and trust HQ hooks for Codex, Grok, and Claude Code
Aliases: master-sync
hq reindex [options]
Option
Description
--repo-root <path>
HQ root to operate on (defaults to the current directory)
--from-hook
Invoked from a Claude/Codex lifecycle hook: never wait on the per-root operation lock — if a sync/rescue holds it, skip this reindex instead of blocking the session (equivalent to —lock-timeout 0)
--lock-timeout <seconds>
Bound the wait for the per-root operation lock (seconds). 0 = refuse immediately; omitted = wait indefinitely (the interactive default). —from-hook implies 0; an explicit —lock-timeout wins.
hq rescue
Re-sync your HQ core to the latest release, preserving your local edits (drift)
hq rescue [options]
Option
Description
--hq-root <path>
HQ root to operate on (defaults to auto-detected root)
Use the staging channel (indigoai-us/hq-core-staging@main)
--floor-sha <sha>
Pin the three-way history floor to a 40-char commit SHA
--paths <list>
Comma-separated top-level paths to narrow the rescue to
--check
Plan only — classify and report, change nothing on disk (—dry-run)
-y, --yes
Skip the confirmation prompt
--no-backup
Skip the pre-op safety snapshot under ~/.hq/backups
--no-self-update
Skip updating hq-cli itself to the latest version first
--cloud-update
Cloud-update mode
hq mcp
MCP pack observability (read-only) across Claude + Codex
hq mcp [options] [command]
Subcommand
Description
serve [options]
Run the HQ stdio MCP server; every tool call is answered by hqd (hq daemon status)
login [options]
Sign in to the hosted HQ MCP connector with browser OAuth
status [options]
Show installed MCP packs/servers and their health in BOTH Claude and Codex (PARTIAL/DRIFT flagged; secrets redacted)
capabilities [options]
Advertise that this hq-cli supports contributes.mcp (dual-runtime Claude+Codex registration)
hq repo
Register code repos with HQ (repos/ and companies/manifest.yaml)
hq repo [options] [command]
Subcommand
Description
add [options] <path>
Link a git checkout into repos/<private|public>/<name> and list it under companies.<slug>.repos in companies/manifest.yaml. A checkout already directly under repos/private or repos/public is registered in place.
hq crm
Read and upsert canonical CRM entities in a company vault
hq crm [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
entity [options] [command]
CRM entity operations
entity list [options]
List one CRM entity page as JSON; continue with nextCursor until null
entity get [options] <id>
Read a canonical CRM entity as JSON
entity upsert [options]
Create or update a canonical CRM entity (POST /crm/entities). Use —type/—name (+ optional external-id flags) or —json <file|-> to pass a full entity/array.
hq company
Company-level settings
hq company [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
transfer [options] [command]
Transfer company ownership. Two-party: initiate nominates, and ownership moves only when the nominee runs accept.
transfer initiate [options]
Nominate an active member as the new owner. Does NOT transfer ownership — the nominee must accept. Prompts for confirmation unless —yes.
transfer accept [options]
Accept a pending nomination and become the owner. This is the step that actually moves ownership. Prompts for confirmation unless —yes.
transfer decline [options]
Decline a pending nomination (nominee only).
transfer cancel [options]
Cancel a nomination you initiated (initiating owner only).
transfer status [options]
Show the current nomination and the company’s transfer history.
rename [options] <new-name>
Rename a cloud-backed company display name
settings [options] [command]
Per-company settings (owner-only)
settings set [options]
Set per-company settings (PUT /company-settings). Require at least one of —crm-enabled / —ontology-enabled / —accounting-enabled / —checkpoint-gate-enabled / —work-registry-enabled / —work-mesh-live-enabled.
settings get [options]
Show whether the checkpoint Stop gate is on for this company (GET /company-settings/checkpoint-gate; any member may read it) and, for an owner or admin, the two work tracking switches (GET /company-settings).
hq agents
Manage a company’s cloud (fleet) agents
hq agents [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
models [options]
List the model catalog available to cloud agents
message [options] <agent> <message...>
Send a message to an agent and wait for its reply
thread [options] <agent>
Show your recent conversation with an agent
terminal [options] <agent>
Open an interactive terminal on an agent’s box — a real TTY over SSM Session Manager (no SSH, no local AWS credentials). Owner/admin only; requires the AWS session-manager-plugin installed locally. With —forward, tunnels a port on the box instead.
provision [options] <name>
Provision a new cloud agent (company-specific monthly price shown before creation)
login-code [options] <agentUid> [code]
Submit a Claude subscription sign-in code for an agent
list [options]
List the company’s agents
status [options] <agent>
Show an agent’s setup state and runtime detail
rename [options] <agentUid> <name>
Set an agent’s display name
set [options] <agentUid>
Update an agent’s profile (name / title / description)
config [options] <agentUid>
Update an agent’s runtime config (model / reasoning effort / service tier), or migrate its brain/runtime provider with —provider (DESTRUCTIVE: terminates + reprovisions the box)
start [options] <agentUid>
Start an agent’s box
stop [options] <agentUid>
Stop an agent’s box
retry [options] <agentUid>
Resume an agent’s setup from the first non-done step
rm [options] <agentUid>
Deprovision (permanently tear down) an agent
rotate [options] <agentUid>
Issue a fresh enrollment code for an external agent (old secret + host key stop minting)
revoke [options] <agentUid>
Revoke an external agent: disable its login and drop it from the team
jobs [options] [command]
List, pause, resume, or cancel an agent’s scheduled jobs
jobs list [options] <agentUid>
List an agent’s scheduled jobs
jobs pause [options] <agentUid> <jobId>
Pause a job’s schedule (reversible)
jobs resume [options] <agentUid> <jobId>
Resume a paused job’s schedule
jobs cancel [options] <agentUid> <jobId>
Cancel a job and delete its schedule
hq agent
Enroll and run this host as an external HQ agent
hq agent [options] [command]
Subcommand
Description
enroll [options] <code>
Enroll this host as an external HQ agent using a one-time code
kit [options] [command]
Install, inspect, or run the external-agent background services
kit install [options]
Install the inbox and heartbeat services plus the skills directory
kit status [options]
Show whether each kit service is installed and running
kit uninstall [options]
Stop and remove the kit services (credentials and skills are kept)
kit wake [options] [command]
Wake this bot’s own framework when HQ messages arrive (webhook URL or local command)
kit run [options] <service>
Run one kit service in the foreground (sync|inbox|heartbeat), or all to supervise every service in one process
probe [options]
Verify identity, vault access, presence, DMs and secrets; report to the console
mcp [options]
Serve HQ as a stdio MCP server for the bot framework (whoami, search, files, secrets-exec, dm, inbox, work-mesh)
inbox [options] [command]
List this agent’s pending HQ messages (mirrored by the kit), or mark them handled
inbox done [options] <ids...>
Mark inbox items handled and ack them on the server
hq outposts
Manage personal and company HQ Outposts (EC2 boxes)
hq outposts [options] [command]
Subcommand
Description
provision [options]
Provision a new Outpost ($80/month — requires —yes)
list [options]
List every Outpost you own, or a company’s Outposts with —company
status [options]
Show live detail for one Outpost
jobs [options] [command]
Register, edit, pause, resume, delete, or run a job on a company Outpost
jobs register [options]
Register a company job and print the accepted revision
jobs edit [options] <jobId>
Change a company job. Omitted fields stay as they are. Requires —revision
jobs pause [options] <jobId>
Pause a company job
jobs resume [options] <jobId>
Resume a paused company job
jobs delete [options] <jobId>
Delete a company job (requires —yes)
jobs run-now [options] <jobId>
Run a company job once now
jobs runs [options] <jobId>
List recent runs of a company job and their outcomes (owner or admin)
exec [options] <command...>
Run a shell command on an Outpost and print its output (use — before flags meant for the remote command). The command waits for the submitted job to finish. Use —wait-timeout-seconds to bound local waiting, or —detach to print the commandId and return immediately.
exec-stage [options]
Stage a file for an asynchronous Outpost command
exec-submit [options] <command...>
Submit an asynchronous shell command to an Outpost (returns immediately with commandId; shell budget defaults to 48h)
exec-result [options]
Fetch the result of an asynchronous Outpost command
Download one file from an Outpost through a short-lived download link (for files too large for exec output). <outpost> is the Outpost id (primary for your first box). The remote path must be inside the Outpost user’s home or HQ dir; relative and ~ paths start at the home dir. The file is saved to [local-path], or to the remote file name in the current directory.
terminal [options]
Open an interactive terminal on an Outpost — a real TTY over SSM Session Manager (no SSH, no local AWS credentials). Requires the AWS session-manager-plugin installed locally. With —forward, tunnels a port on the box to your machine instead of opening a shell.
refresh-config [options]
Re-apply the server’s job runner configuration to the company Outpost (owner or admin)
codex-enable [options]
Enable (or retry) Codex on an Outpost
login [options]
Request a fresh login URL for an Outpost
login-code [options] <code>
Submit the Claude sign-in code for an Outpost that is awaiting login
destroy [options]
Tear down (permanently destroy) an Outpost
hq billing
Inspect billing and mint a card-capture link
hq billing [options] [command]
Option
Description
--company <slug>
Company slug (resolves to companyUid)
Subcommand
Description
status [options]
Show subscription status and whether a card is on file
checkout [options]
Mint a Stripe card-capture link (shareable) to add a card
upgrade [options]
Open the switch to usage pricing for a company on an earlier plan (usage-pricing companies have nothing to change)
sponsor [options] [command]
Ask a company to cover your personal HQ plan, or answer a request
sponsor request [options] <company>
Ask a company to cover your personal HQ plan
sponsor list [options]
Show sponsorship requests — the company’s queue, or your own
sponsor approve [options] <id>
Say yes — put someone’s personal plan on the company bill
sponsor decline [options] <id>
Say no to a plan sponsorship request
sponsor revoke [options] <id>
End a sponsorship the company is currently paying for
sponsor withdraw [options] <id>
Take back a sponsorship request you made
sponsor grant [options] [person]
Admin: put a member’s personal plan on the company bill, no request needed
sponsor ungrant [options] <person>
Admin: take a member’s personal plan off the company bill
usage [options]
Show this month’s usage, projected bill, included usage left and what drives it (read-only)
hq vault
Manage HQ vault settings
hq vault [options] [command]
Subcommand
Description
retention [options] [command]
Get or set the default retention for future vault object versions
retention get [options]
Show the vault’s default retention in days
retention set [options] <days>
Set the vault’s default retention in days (1-30)
hq search
Search the local HQ qmd index
hq search [options] [command] [query...]
Option
Description
--mode <mode>
Search mode: keyword, semantic, or hybrid Default: "keyword".
-c, --collection <collection>
Restrict search to a collection
-n, --count <count>
Maximum result count
--json
Request machine-readable qmd output
Subcommand
Description
get [options] <document>
Retrieve a qmd document by path or document id
hq library
Company media library (.hqcloud asset placeholders)
hq library [options] [command]
Subcommand
Description
search [options] [query...]
Search library assets in the cloud, falling back to the local index when offline
add [options] <sources...>
Upload files or URLs to the company vault and add .hqcloud placeholders to the library
tag [options] <placeholder>
Add or remove tags on a library placeholder
ls [options] [dir]
List library placeholders in a folder (default: companies/)
describe [options] <placeholder>
Re-run the AI description and suggested tags for a library image or video
fetch [options] <placeholder>
Download a library asset next to its placeholder (wraps hq storage fetch)
link [options] <target>
Mint a short-lived download URL for a library asset (needs files:read)
rm [options] <targets...>
Remove library assets: the local original (if fetched), the placeholder, and its vault copy. Keeps the shared blob unless —purge. Previews and asks first unless —yes
hq index
Manage the local HQ search index
hq index [options] [command]
Subcommand
Description
sync [options]
Reconcile collections and incrementally update the qmd index
collections [options]
Show expected and registered qmd collections
background [options]
Run a detached, single-flight qmd cleanup and reindex
status [options]
Show qmd binary, collection, and index status
hq capacity
Show provider capacity from the local ledger (read-only; enable, login, and ingest write to it)
hq capacity [options] [command]
Option
Description
--json
Print now.json plus schemaVersion
--widget
Print an HTML usage card for chat
Subcommand
Description
history [options]
Print readings from the ledger event log
doctor [options]
Report last success, pinned versus installed CLI versions, and parser failures
route [options]
Pick the account with the most capacity for a new lane (read-only; port of boss-route)
enable [options]
Create the ledger directory and mark the collector on
disable [options]
Mark the collector off without deleting readings
ingest [options] [command]
Record a provider capacity reading
ingest claude-statusline [options]
Read one Claude Code statusline JSON payload from stdin and record its rate limits
ingest refusal [options]
Record a classified provider rate-limit refusal as an exhausted reading
login [options] [provider]
Show signed-in accounts, or sign into a provider through hq lanes accounts login
accounts [options] [command]
List provider accounts from the registry and default homes
Rename a registry account id. Does not touch the provider credential
accounts remove [options] <provider> <id>
Forget a registry account. Does not delete its home unless —delete-home is set
hq doctor
Verify HQ hook wiring, runtimes, and connection health (read-only; integrations uses one control-plane inventory read).
hq doctor [options]
Option
Description
--json
Emit the machine-readable JSON document (no colour).
--verbose
Also print every PASS result in text output.
--no-color
Disable ANSI colour even on a TTY.
--session-id <id>
Scope the runtime probe’s ledger check to this exact session.
--deep-test
Also fire pure-guard hooks through the real gate under all three profiles (sandboxed).
--live-runtimes
Also probe each installed AI CLI (claude, codex, grok) with a one-line prompt to verify login and subscription (networked; uses your subscriptions).
--company <slug>
Company slug for integrations checks (otherwise resolves your single active company).
--only <families>
Run only these comma-separated check families (e.g. hooks). Unknown names are rejected; —json then records the scope.
--fix
Apply the allowlisted safe repairs (backs up first; read-only without this flag).
--sweep-temp-homes
Delete idle temp homes (90+ minutes untouched, not in use): tmp.* folders holding .rustup or .cargo, and hq-temp-home.* folders left by the wrapper. Reports GB freed.
--dry-run
With —sweep-temp-homes, list what would be removed and delete nothing.
--yes
Skip the interactive —fix confirmation (non-interactive use).
--force
Let —fix run despite uncommitted changes under .claude/, .codex/, or .grok/.
hq mesh
Work mesh — register project work, report progress, and warm ~/.hq/work-mesh/cache
hq mesh [options] [command]
Subcommand
Description
check [options]
Show active work-mesh threads for a company/project
start [options]
Ensure a project thread exists and claim/report start
progress [options]
Append a progress event to the project thread
blocked [options]
Append a blocked event to the project thread
done [options]
Mark the project thread done
note [options]
Append a note with no status change
story [options]
PATCH one Board story status and/or assignee
claim [options]
Assign a story to yourself and move it to in_progress
overlap [options]
List other sessions working on the same story, branch, or files
suggest [options]
Rank teammates and agents to pick up a story
roster [options]
Show each member’s daemon and session reporting state
project [options] [command]
Set a project Board view or register its project channel
project set [options] <projectId>
PUT the project Board view from flags or a PRD JSON file
project ensure [options] <projectId>
Idempotently register a PRD on the Board and verify the view
project reconcile [options] <projectId>
Read the project Board and report drift from a PRD ledger
project register [options] <projectId>
Register an existing Board and bind its project channel; use project ensure to create or repair a Board
project reassign [options] <projectId>
Move project ownership to another person or agent
project archive [options] <projectId>
Archive a Board (for example a duplicate); keeps stories, folder, channel, members and history
project complete [options] <projectId>
Mark a Board completed; it leaves the active list and organize candidates
project activate [options] <projectId>
Restore a completed or archived Board to active
project lifecycle [options] [projectId]
List completed and archived Boards, or show one project’s lifecycle status
project member [options] [command]
Manage project members
doctor [options]
Warm ~/.hq/work-mesh/cache from hq-pro (directory, inbox, pair DMs). Not hq doctor.
context [options] [command]
Work context reconcile, organize, correct, device default, and untracked consent
context reconcile [options]
Resolve session identity and company; durable local accept before network
context resolve [options]
Interactively ask (once per repo) which company this repo’s work is filed under, and remember it (gap 4). Person TTYs only — never agent boxes / non-interactive.
context backfill-held [options]
Reconcile ENDED sessions whose held events lack a company so the daemon re-attributes the backlog (explicit, opt-in; no fleet fan-out)
context bind-project [options] <target> [slug]
Bind this session to a project id, a numbered candidate, or new <slug>
context prd-sync [options]
Emit task_status for PRD story passes/status changes on a bound session
context auto-bind [options]
Bind a needs_project session to the project its repo and branch belong to (local-only, fail-open)
context organize [options] [action]
List or submit one-time project/task decisions (US-007B / US-005B)
context correct [options]
Explicit cross-company session correction via server migrate (US-017B)
context requeue [options]
Requeue quarantined outbox operations (default: AUTH_DENIED - work-mesh-live gap 10 recovery - plus sessions refused while work tracking was off). Run once after the company owner turns work tracking on
context default [options] [command]
Manage the device-local default company preference
context untracked [options] <sessionId>
Record an explicit user-origin untracked consent receipt (local only)
contract [options] [command]
Print versioned work-mesh contracts for other repositories
Local session-event spool: enqueue (printf-compatible), flush, and live status
session start [options]
Enqueue session_start
session turn-start [options]
Enqueue turn_start
session turn-end [options]
Enqueue turn_end
session end [options]
Enqueue session_end
session task-status [options]
Enqueue task_status
session blocked [options]
Enqueue blocked
session note [options]
Enqueue note
session flush [options]
Claim spool/held by rename, hold or drop by context state, POST batches of ≤100
session status [options]
Print the company-wide live read (US-004) as a table or —json
mode [options] [action]
Get/set the daemon emit mode (legacy = spool/flush; direct = receive-only). check probes whether the server route is live.
emit [options]
Direct-emit drain: POST any deferred events (retry file) to /v1/mesh/events; —replay-legacy also replays the legacy spool/held backlog through the new endpoint
daemon [options] [command]
Resident spool drainer + MQTT presence (install as a user service)
daemon run [options]
Start the single-instance daemon (pid lock, spool watch, presence)
daemon install [options]
Install LaunchAgent (macOS), systemd user unit (Linux), or Scheduled Task (Windows)
daemon uninstall [options]
Remove the installed user-level daemon service
daemon status [options]
Show whether the daemon service is installed/loaded
daemon doctor [options]
Report daemon running/MQTT/spool/held/dead-letter/outbox/flush health
hq lanes
Create, supervise, and steer worker lanes; routed decision queue
hq lanes [options] [command]
Subcommand
Description
link [options] [command]
Link live sessions to a conductor through session inboxes
link open [options]
link join [options]
link send [options]
link ack [options]
link capabilities [options]
link report [options]
link status [options]
link inbox [options]
link read [options]
link wait [options]
link list [options]
link leave [options]
link close [options]
reap [options]
Classify local lanes whose session senior has ended; report only by default
checks [options]
Evaluate tracker and lane health checks
orphans [options]
Report lanes whose senior has ended or gone idle
repair-dropbox [options] <ulid>
Inspect or redeliver messages stranded in a bare-ULID lane dropbox
report [options]
Generate the owner report for tracker teams
team [options] <slug>
Show a team’s OKRs, KPIs, work, blockers, and questions
hold [options] [command]
Set and enforce repository merge holds
hold set [options] <repo>
Set or replace a repository merge hold
hold list [options]
List active and expired repository merge holds
hold show [options] <repo>
Show a repository merge hold
hold release [options] <repo>
Release an existing repository merge hold
hold check [options]
Refuse merge commands targeting an actively held repository
review [options] [command]
Record manager review of pull-request heads
review record [options] <repo> <pr-number>
Record a review against an exact pull-request head
review show [options] <repo> <pr-number>
Show the review record for one pull request
review list [options]
List recorded pull-request reviews
outpost-pruner [options] [command]
Inspect, install, and run safe outpost worktree pruning
outpost-pruner run [options]
Run the guarded outpost worktree and user-cache cleanup
outpost-pruner install [options]
Write the pressure and daily systemd user timers
outpost-pruner uninstall [options]
Disable and remove the pressure and daily systemd user timers
kpis [options] <lane>
Show runtime freshness for a manager lane’s key results
keep-alive-audit [options]
Audit lead lane keep-alive timing. Durable worker-resume events retain launch time and trigger after bounded lane history rolls over; late_restarts_after_window counts known launches after the grace window.
autopilot [options] [command]
Run lead autopilot checks and report follow-through metrics
autopilot report [options]
Report operator nudges, review latency, and idle slots with ready work
autopilot run [options]
Run one lead autopilot check pass
create [options]
Register a lane, run admission, and return immediately
progress [options] <lane-id>
Report on a subagent lane: a progress line, the final envelope, or a failure
recover-loop [options] <lane-id>
Preview or explicitly fail/abandon uncertain loop effects; never replay work
enqueue [options] <lane-id>
Validate and queue one phase envelope for a loop lane
keep-alive [options] <lane-id>
Opt a lane into or out of automatic lead-lane resumes
set [options] <lane-id>
Change lane settings or bind its missing company attribution
grant [options] <lane-id>
Set a lane’s alive child-lane grant as its senior
backfill-grants [options]
Preview missing hire grants for alive manager lanes; writes are disabled
doctor [options]
Report active keep-alive lanes blocked by retired, superseded, or pending-question state
list [options]
List lanes
status [options]
Summarize project stories, lanes, PRs, blockers, and pending questions
capacity [options]
Report this host’s lane capacity by provider and account
outpost-identity [options] [command]
Repair this host’s Outpost identity marker
outpost-identity backfill [options]
Write a missing Outpost identity marker without replacing an existing value
move [options] <lane-id>
Move an unstarted queued lane to a host with capacity
admit [options] <lane-id>
Try to admit the oldest eligible queued lane
wait [options]
Wait for the first requested change across lanes
watch [options] <lane-id...>
Watch lane state and question changes until all lanes are terminal
monitor-check [options]
Check that active Claude senior lanes have live Monitor watch tasks
show [options] <lane-id>
Show one lane
story [options] [command]
Track several project stories on one lane
story list [options] <lane-id>
List the stories carried by a lane
story add [options] <lane-id> <story-id>
Add and start another story on a lane
story start [options] <lane-id> <story-id>
Start or resume a carried story
story review [options] <lane-id> <story-id>
Move a carried story to review
story block [options] <lane-id> <story-id>
Mark one carried story blocked
story done [options] <lane-id> <story-id>
Mark a reviewed story done after senior verification
message [options] <lane-id>
Send a follow-up message to a lane
resume [options] <lane-id>
Resume an interrupted, done or blocked lane, or an awaiting_input lane with no answerable question; recover failed lanes with —from-failed as a senior ancestor or same-company manager
reparent [options] <lane-id>
Rewrite a lane’s senior edge
repair-company [options] <lane-id>
Set missing company attribution on a legacy lane
logs [options] <lane-id>
Print a lane’s local logs
stop [options] <lane-id>
Cancel a queued lane or kill a running one (terminal; use interrupt to keep it resumable)
interrupt [options] <lane-id>
Stop a running lane’s turn and keep it resumable with hq lanes resume
approve [options] <lane-id>
Record senior approval for a registry worker checkpoint
ask [options]
Queue a mid-run question without blocking the lane
questions [options] [command]
Routed decision queue
questions list [options]
List questions
questions show [options] <question-id>
Show one question
questions answer [options] <question-id>
Answer a pending question
questions escalate [options] <question-id>
Advance a question one hop
questions cancel [options] <question-id>
Cancel a pending question
accounts [options] [command]
Provider account registry (configured vs logged in)
accounts list [options]
List accounts
accounts balance [options]
Report Claude account load and suggest safe between-turn moves
Compress or remove unreferenced old Codex session rollouts
hq monitor
Run detached monitors and deliver their output into agent context
hq monitor [options] [command]
Subcommand
Description
start [options] [argv...]
Start a detached monitor script
list [options]
List active monitors
show [options] <monitor-id>
Show monitor metadata, state and events
stop [options] <monitor-id>
Stop a running monitor
logs [options] <monitor-id>
Read monitor events or stderr
hq bot
Personal local bots that run on this computer with your own model login
hq bot [options] [command]
Subcommand
Description
continuity-import [options] <snapshot>
Import verified private bot context before cloud activation (promotion coordinator)
jobs [options] <name>
Show, schedule, or remove a bot’s recurring jobs (the schedule runs in the cloud, not on this computer)
promote [options] <name>
Move this local bot to cloud compute, preserving its identity and private context
create [options] <name>
Provision a local bot identity, scaffold its worker folder, install its launchd agent, and start it
companies [options] <name>
Show or change which companies a company bot is a member of (you invite or remove it; a personal bot has no list of its own)
workers [options]
List the workers (templates) a bot can be created from (hq bot create <name> —worker <id>)
list [options]
List local bots with runtime, online state, and pid
adopt [options] <name>
Bring a bot you already own back to this computer: new machine credentials, its saved settings, worker folder, and startup agent
restore [options]
Bring back every local bot your HQ account owns that is not set up on this computer (after a reinstall, or on a new Mac)
start [options] <name>
Start the bot (via launchd when installed, else a detached process)
stop [options] <name>
Stop the bot process (it starts again at next login while its launchd agent is installed)
restart [options] <name>
Stop then start the bot
status [options] <name>
Show the bot’s process, heartbeat, launchd, and credential state
auth [options] <name>
Sign a runtime-backed personal bot in to ChatGPT, Grok, Claude, or another supported provider
logs [options] <name>
Show the bot’s log
run [options] <name>
Run the bot in the foreground (this is what launchd runs)
rm [options] <name>
Stop the bot, remove its launchd agent, delete its cloud identity and local state
set [options] <name>
Change the model and thinking level a bot uses (applies from its next message), or the name people see for it (saved in HQ; a running bot restarts to use it)
set-model [options] <name> <model>
Set the model this bot asks its runtime for (stored in the bot’s own settings; the runtime’s own config is never changed), or default to clear it
probe [options]
Run one minimal test turn with a runtime, model and thinking level exactly as a bot would, and say whether a bot set up that way can answer. Costs one tiny model turn on your own login; runs in an empty scratch folder, so no HQ settings or hooks load
intro [options] <name>
Have the bot re-send its introduction DM to you
daemon [options] [command]
Manage the bot’s launchd user agent (macOS)
daemon install [options] <name>
Install and load the LaunchAgent so the bot starts at login and restarts on crash
daemon uninstall [options] <name>
Unload and remove the LaunchAgent
daemon status [options] <name>
Show whether the LaunchAgent is installed and loaded
hq report
Publish and manage a channel report owned by this bot.
hq report [options] [command]
Subcommand
Description
get [options] <reportId>
publish [options] <reportId>
promote [options] <reportId>
discard-draft [options] <reportId>
access [options] <reportId>
decommission [options] <reportId>
adopt [options] <reportId>
templates [options]
hq deploy
Manage deployments in hq-deploy
hq deploy [options] [command]
Subcommand
Description
comments [options] [command]
List and read comments on your deployed apps
comments list [options] <app>
List comments on an app in your org
comments read [options] <app> <commentId>
Read a comment on an app in your org
list [options] [query]
List deploy apps in your company, optionally filtered by name or subdomain
access [options] [command] [app]
Show or change who can open a deploy app (company, selected members, or public)
access get [options] <app>
Show an app’s current access mode and selected members
domains [options] [command]
Manage custom domains on your deployed apps, and buy domains
domains list [options]
List the custom domains on an app
domains add [options] <domain>
Link a custom domain to an app and print the DNS records to create
domains refresh [options] <domain>
Check DNS validation and start serving once the certificate is issued (exits 2 while not live)
domains remove [options] <domain>
Unlink a custom domain from an app
domains base [options] [command]
Manage the company deploy domain: one domain for every app (<app>.<domain>)
domains search [options] <query>
Check whether a domain can be bought, and its price
