staqly CLI

The staqly command writes, schedules and publishes posts, connects channels, manages schedules and watches the queue, all from a terminal or a script. The same binary also runs Staqly Community Edition, the whole server, with staqly serve.

Pick the CLI for shell scripts, cron jobs and quick checks. Use an SDK inside a TypeScript or Python program, MCP when an AI agent such as Claude Code talks to staqly directly, and the REST API from any other language.

Get startedCopy Anchor Link

curl -fsSL https://raw.githubusercontent.com/staqly/staqly/main/scripts/install.sh | sh

staqly version

staqly serve

staqly channels connect custom --name "My publisher"      # your own automation; see `staqly channels providers` for the rest
staqly collections create "Product updates"
staqly posts create "Product updates" --content "v1.4 is out." --channel "My publisher" --publish-at "+1h"

staqly schedule                 # the calendar for the next 7 days
staqly posts get post_oalirbz67y4m4cqxq5dw
staqly jobs list --status failed

Other ways to installCopy Anchor Link

Windows. Download the .zip for your CPU from the releases page and put staqly.exe on your PATH.

A specific version or folder. The install script reads two environment variables:

curl -fsSL https://raw.githubusercontent.com/staqly/staqly/main/scripts/install.sh | STAQLY_VERSION=v0.2.0 sh

From a checkout of the repository. Requires Go and Node:

make dashboard   # embeds the dashboard for `staqly serve`
cd apps/cli
make build       # → apps/cli/bin/staqly
make install     # → $GOBIN, or ~/go/bin

Shell completion:

source <(staqly completion bash)        # or zsh
staqly completion fish > ~/.config/fish/completions/staqly.fish

Run the serverCopy Anchor Link

staqly serve runs the API, MCP endpoint, job queue and dashboard in one process.

staqly serve                                         # dashboard on this computer only
staqly serve --expose                                # dashboard reachable from phones and tablets
staqly serve --data ~/staqly-data --port 4333 --dashboard-port 9090

With a subcommand, staqly serve runs an operator task against the same data instead: token [--rotate], connect-link, backup <file> and storage migrate. See Run the binary and Operator CLI.

Connect to a serverCopy Anchor Link

By default the CLI talks to Staqly Community Edition at http://localhost:3333, which wants its access token: against staqly serve on this computer the CLI reads it from the data folder (STAQLY_DATA_DIR, else ~/.local/share/staqly) when no key is set. Started it with --data somewhere else? Set STAQLY_DATA_DIR to that folder too. Against Docker pass the token as a key. To use another server, pass flags, set environment variables, or save a profile.

The first one set wins, in this order: flag, environment variable, selected profile, default.

Save servers as profilesCopy Anchor Link

staqly login --key "$(docker exec staqly server token)"        # Community Edition in Docker → profile "default"
staqly login --url https://app.staqly.ai --key stk_... --profile work
echo "$KEY" | staqly login --url https://app.staqly.ai --key - --profile ci
staqly profiles                 # list (keys masked); * marks the current one
staqly profiles use work        # switch
staqly status --profile ci      # use a profile for one command
staqly logout --profile ci      # remove a profile

login checks the credentials against GET /v1/session before saving (skip with --no-verify), writes the config file (folder 0700, file 0600) and makes the profile current. Without --key it uses STAQLY_API_KEY or asks for the key at a hidden prompt.

How keys are protectedCopy Anchor Link

• A key stored in a profile is only sent to that profile's own URL. If --api-url or STAQLY_API_URL points somewhere else, the CLI sends no key and prints a note. Pass --key or STAQLY_API_KEY to authenticate against the other server.
• The CLI warns when a key would travel over plain http:// to a host other than localhost. Use https:// for remote servers.
• Channel credentials passed with --field go straight to the API and are never written to the config file.

Output formatsCopy Anchor Link

• Human-readable tables by default. Colors only appear on a terminal; NO_COLOR=1, TERM=dumb or --no-color turn them off.
• --json prints the API response unchanged. Streaming commands (events tail --follow) print one JSON object per line (NDJSON).
• With --json, errors go to stderr as {"error":{"code","status","message","fields","details","retry_after_seconds"}}.
• Text from the server (post content, channel names, error messages) is treated as untrusted. Human output strips control characters, ANSI / OSC escape sequences and Unicode bidi overrides before printing. --json keeps the data intact but writes C1 controls and bidi overrides as \uXXXX escapes, so it is still safe to view in a terminal.

Exit codesCopy Anchor Link

Error messagesCopy Anchor Link

The CLI turns API errors into plain-language hints:

When the server can't be reached, the error names the URL it tried.

Next stepsCopy Anchor Link