Reference
Every command, straight from the binary.
Generated from the CLI’s own help output in 0.4.2. If a flag changes and this page does not, the build fails.
SETUP
setup --space <id> --actor <label>- persist one explicit human space + attribution
use [<space>|--clear]- bind/show this directory's space; keep global default
onboard --space <id> --actor <label> --json- create/join a space + return executable recipes
demo [--json]- isolated add → list → done journey; no live data
init- create ~/.tasq/db.sqlite + config
config [show|get|set <k> <v>] manage ~/.tasq/config.jsonfeedback "summary"- capture actionable context locally, even offline
AREAS
area list- list all areas
area show <slug>- show area details
area add <name> --slug <s> [--importance 1-5] [--cadence <text>]area update <slug> [...]area delete <slug>
GOALS
goal list [--area <slug>] [--status active|paused|done|abandoned]goal add <title> --area <slug> [--horizon <text>] [--importance 1-5]goal update <id> [--status ...] [--horizon ...] [--importance ...]
PROJECTS
project list [--status ...] [--goal <id>]project add <title> [--goal <id>] [--area <slug>]project update <id> [--status ...]
TASKS — the core verbs
add <title> [--area <slug>] [--goal <id>] [--project <id>]- [--next <text>] [--due <iso>] [--est <min>] [--priority 1-5] [--recurrence daily|weekly|monthly|yearly] [--interval N] [--anchor due|scheduled|completion] [--success <criteria>] [--completion assertion|evidence] [--idempotency-key <key>] --completion evidence requires --success; --validated requires --completion evidence
list [--status ...] [--area <slug>] [--goal <id>] [--project <id>]- [--priority 1-5]
show <id>inspect <id> [--json]- canonical commitment graph + resume cursor
update <id> [--title ...] [--next ...] [--due ...] [--recurrence ...] [...]start <id> [--note <text>]done <id> [--evidence <id,...>] [--note <text>] [--source <text>] [--at <iso>]block <id> --reason <text>unblock <id>cancel <id> [--reason <text>]reopen <id>delete <id>restore <id>
DEPENDENCIES
depend <id> --on <other-id> [--type blocks|discovered_from|relates_to|duplicates]- record that <id> depends on <other-id>
undepend <id> --on <other-id> [--type ...]- remove a dependency edge
capture <id> <title> [--context <json>] [--idempotency-key <key>]- atomically file linked work; keep current claim
WHY THE WORK EXISTS
add <title> --because "<one sentence>"- record what this work rests on; several tasks can rest on the same sentence
wrong "<sentence>" --reason <text> [--evidence <id,...>]- that turned out false: pauses every task resting on it. Nothing is cancelled.
why <id>- the chain: what this rests on, who stated it, who withdrew it and on what evidence
resume <id> --reason <text>- continue a paused task anyway; unlinks it from the withdrawn sentence and records why
because list [--status standing|withdrawn]- what this workspace currently believes
because attach <id> "<sentence>"- bind an existing task to a sentence
AGENT COORDINATION
agent install codex|claude|generic --space <id> --actor <label>- preview an exact host-bound MCP registration
agent instructions --space <id> [--write|--check]- manage the versioned Tasq block in AGENTS.md
fleet- who is holding what right now, and for how much longer their lease runs
claim <id> [--for 30m] [--force]- atomically claim work (repeat to heartbeat); --force takes work whose blockers are unresolved
release <id>- release the current claim
attempt start <id> [...]- record one concrete execution
attempt succeed|fail <id>- close an execution attempt
evidence add <id> --kind ...- attach an observable receipt
evidence list [<id>]- inspect completion evidence
cost budget|record|show ...- bound and inspect observed attempt cost
signature show <id>- inspect exact accepted signed proof
signature bindings [record]- inspect typed proof bindings (read-only)
resolution contract|trust|propose|challenge|attest|settle|adjudicate|show- independently validate completion
premise show|propose|challenge|decide- inspect or refute why a commitment exists
resource acquire|renew|release|verify|get|list|events|sweep- coordinate any opaque external resource key
mcp --tenant <space> --actor <label> [--capabilities read,coordinate]- [--completion assertion|evidence] run a capability-scoped local MCP stdio server
web --tenant <space> [--host 127.0.0.1] [--port 4137]- explicit foreground read-only Local Console
web status --tenant <space> --json- prove a registered Console listener is live
REMOTE SERVER
remote enroll --endpoint <https-url> --workspace <id> [--profile <name>]- redeem TASQ_ENROLLMENT_TOKEN into a private profile
remote status [--profile <name>]remote list|show|events|operations [--profile <name>] [--json]remote call <operation> --resource-kind <kind> --resource-id <id>- --input <json> --idempotency-key <key> [--expected-revision <n>]
remote logout [--profile <name>]- remove local credentials; server revocation is separate
WAIT / OBSERVE / RECONCILE
wait create <task> --kind <kind> --parameters <json> [--deadline <iso>]wait list [task]- inspect typed external expectations
wait cancel <wait> --reason <text>wait sweep [--at <iso>]- reconcile queued facts, then expire due waits
observation ingest --source <s> --external-event-id <id>- --kind <kind> --payload <json> --occurred-at <iso>
observation list|show [...]- inspect immutable normalized facts
reconcile <wait> <observation>- run the frozen deterministic matcher
reconcile list [wait]- inspect immutable decisions and effects
MACHINE DISCOVERY
discover [--json]- capabilities, extensions, schemas + cursors
discover schema <resource-id>- fetch one digest-bound JSON Schema
discover negotiate --hello <json>- strict cold-start compatibility handshake
VIEWS
context [--max-records N] [--max-tokens N] [--json]- bounded reason-traced universal state packet
summary add|list|show [...]- source-bound compact context for closed work
context-link attach|detach|list|show [...]- reusable external context pointers; no content
next [--limit N] [--area <slug>] [--priority 1-5]- prioritized next-action list
search "<query>"- substring search across task text
inbox- tasks without project
tree <id>- show a task + its sub-tasks
task status <id>- progress + ETA for a task w/ sub-tasks
project status <id>- progress + ETA for a project
AUDIT
event list [--since <iso>] [--entity-id <id>] [--actor <name>]- [--entity-type area|goal|project|task] [--after-sequence N] [--ascending]
PROJECTION
projection [--target <path>]- regenerate markdown projection (writes to config.projectionTarget if no --target)
STORE SAFETY
store status- what format this store is, and whether opening it would apply an irreversible upgrade
store upgrade- apply a pending format upgrade deliberately; a verified snapshot is written first
store recovery-points- verified snapshots written before each upgrade
store restore <id> [--force]- roll back to one, after checking its digest
store clone --to <dir>- an independent copy to work on: WAL-safe, with every path rewritten inside the copy
DURABILITY
backup [<path>] [--rotate N]- snapshot DB to ~/.tasq/snapshots/db-<ts>.sqlite (keeps last N snapshots if --rotate is set)
export [<path>]- bounded portable workspace export (not a backup)
import <export.json> --db <new-db-path>- validate fully, then create one new store
doctor [--fix-permissions] [--repair-outbox]- verify/repair delivery, journal and private modes
journal checkpoint --accept-database --reason <text>- archive history and accept the DB cursor baseline
META
version- print version
help [command]- this message
FLAGS
--json / -j- machine-readable JSON output
--actor <name>- override default actor
--tenant <id>- override default tenant (rare)
Agents should not read this page. Run
tasq onboard --space <id> --actor <label> --json and execute the returned recipes, which are versioned by the binary you actually have.