Skip to content
tako
Esc
↑↓navigate↵open⌘Jpreview
On this page

CLI Reference

Look up Tako command groups, global options, environment selection, and structured output for local development, deployment, and operations.

tako [--version] [-v|--verbose] [--ci] [--json] [--dry-run] [-c|--config <CONFIG>] [--ssh-passphrase <PASSPHRASE>] <command>

Progress, prompts, and diagnostics normally go to stderr. Command results and machine-readable data go to stdout. Interactive tako dev owns stdout for its live interface and logs; tako run preserves child stdout.

Use tako <command> --help or the connected MCP tool’s schema for the exact arguments supported by your installed CLI. This reference explains command groups and behavior; task pages cover complete workflows.

Global Options

Option Meaning
--version Print the CLI version and exit.
-v, --verbose Append-only transcript with timestamps, log levels, and debug diagnostics.
--ci Deterministic non-interactive output: no colors, spinners, raw mode, or prompts.
--json Emit structured JSON on stdout. Progress, diagnostics, and errors stay on stderr.
--dry-run Preview supported changes. A command that cannot preview its changes exits with an error before running.
-c, --config <CONFIG> Use an explicit config file. .toml is appended when omitted.
--ssh-passphrase <PASSPHRASE> Use a passphrase for encrypted local SSH keys.

App-scoped commands that honor -c: agents, mcp, init, dev, run, logs, deploy, releases, backups, delete, secrets, credentials, storages, generate, and project-context scale.

Preview changes with --dry-run on agents, deploy, scale, delete, connect, disconnect, ssh, secrets sync, secrets migrate, backups now/download/restore, and workflows retry/cancel. Network membership commands also support previews, including server add, remove, drain, resume, and uninstall. Read-only commands still report their results.

The config file’s parent directory determines the project root and local .tako/ state. Config files in the same directory share secret drafts; a bound environment cannot be reused for another app or network. See alternate configurations.

For finite commands, --json prints one final object. Commands without a specialized schema use {"ok":true,"command":"<command>"}. Failures print {"ok":false,"error":{"message":"..."}} on stdout and the human-readable error on stderr. Specialized objects: status includes servers; doctor includes paths, CA, dev server, apps, and proxy checks; network server list, secrets list, credentials list, dev list, backups list/status, and releases list include their collections; backups download and backups restore include the backup entry and what they wrote. Streaming exceptions keep stdout as JSONL until interrupted: tako logs --tail --json emits one log record per line, and tako dev --json emits a ready object, then log and event records, with no logo or TUI. tako run is also an exception: child stdout stays untouched and no JSON result object is appended.

Installed CLIs send an anonymous event per command so we can count unique users and see which commands run. Set TAKO_TELEMETRY=0 to opt out. See Usage stats.

Project Setup

tako init

Creates tako.toml, updates .gitignore, detects runtime and preset, asks for app name and production route, and installs the SDK package for JS and Go projects. If the production route is a wildcard route, init can collect and encrypt the Cloudflare credential required for Let’s Encrypt DNS-01. Interactive init then offers coding agent setup with a default of yes; non-interactive init leaves agent settings alone.

tako generate
tako gen
tako g

Refreshes generated project files:

  • JS/TS: tako.d.ts with environment names, secret names, storage bindings, channel metadata, typed Store params and data, workflow metadata, and env var names.
  • Go: tako_secrets.go with typed secret accessors.

For JS/TS, generation defaults to the project root and keeps an existing tako.d.ts in app/, src/, or the project root. Empty existing channels/ or workflows/ directories get demo definitions. Declaration files are never scaffolded as handlers.

Generation exports the app’s language-neutral tako-schema.json beside tako.toml. Root-level schemas lists local files or HTTP(S) URLs to import. JavaScript gets generated tako.imports.ts and convenience tako.exports.d.ts beside its existing tako.d.ts; Go gets tako_imports.go, and Rust gets tako_imports.rs. Native outputs use build.cwd or project root. Native exporters use tako-schema/main.go or examples/tako_schema.rs; see Network Primitives for DTO setup and runtime limits. No package exports are edited.

Remote schemas are pinned, with their content, in tako-schema.lock. Commit it for reproducible offline generation. tako gen --update-schemas fetches remote sources again; ordinary generation reuses the lock and rereads local paths. Handwritten tako.imports.ts files must migrate to schemas before generation can replace them.

tako agents

tako agents
tako agents --agent codex,claude-code --yes
tako agents --agent cursor --dry-run --json

Installs bundled workflow and SDK skills and configures the local MCP server for Codex, Claude Code, Cursor, and VS Code Copilot. Detects clients by default; --agent selects them explicitly (repeatable or comma-separated). --yes skips the setup confirmation. --dry-run previews files without writing. All settings are project-local; existing unrelated configuration is preserved. Re-run to refresh after upgrading Tako. See Coding agents.

tako mcp

tako mcp
tako mcp -c /path/to/project/tako.toml

Serves Tako tools, bundled docs, and SDK skills over MCP stdio. Designed to be started by a coding agent; stdout is reserved for MCP protocol messages, including with --json. Tools use existing CLI credentials and permissions. See Coding agents for manual client configuration and tool behavior.

tako dev

tako dev
tako dev vite dev
tako dev --variant preview
tako dev --var preview
tako dev --tunnel
tako dev --restart
tako dev --background --json

Starts or attaches to a local dev session behind trusted HTTPS and real hostnames. --variant foo uses a variant hostname such as myapp-foo.test; --restart restarts the process with the current config instead of attaching. It starts the dev daemon, prepares DNS/proxy/CA setup, generates files, injects secrets and storage through fd 3, waits for fd-4 readiness, and registers routes.

Pass a command after tako dev to override the configured dev command for that run:

tako dev vite dev
tako dev npm run dev
tako dev -- vite dev

Use -- when the child command conflicts with a Tako subcommand such as stop or list, or when the child command starts with a flag.

Interactive controls:

Key Action
l Toggle LAN mode for managed local routes.
t Toggle a temporary public tunnel URL.
r Restart the app.
b Leave the app running in the background and exit the attached CLI.
Ctrl+c Stop the app and exit.

The interactive status panel always shows local routes plus LAN and tunnel state. LAN and tunnel rows include their own enable/disable hints on the same rows; tunnel also shows starting and reconnecting states while it connects. Tunnel hostnames use {app}-{id}.tako.website; the id is derived from the app name and Tako Identity, so the same app gets the same URL when the same identity is available. If the tunnel connection drops, Tako keeps the URL reserved and reconnects automatically. One Tako Identity can have up to five active tunnel URLs connected at the same time; starting a sixth closes the oldest active tunnel for that identity. When a tunnel turns off, tako dev prints a log line with the close reason.

Inactive tunnel URLs show a Tako error page in browsers and machine-readable errors for API clients.

Subcommands:

tako dev stop [name]
tako dev stop --all
tako dev list
tako dev ls

stop without a name stops the app for the selected config file. --all stops all registered dev apps. list shows currently registered dev apps and any active tunnel URLs.

--background exits after registration and readiness while the dev daemon keeps the app running. Combine it with --json for one ready object, then inspect with tako dev list and stop with tako dev stop. Global --json skips the interactive UI. tako dev --json prints a ready object with app, url, and hosts, then JSONL log and event records. tako dev list --json prints one object with an apps array.

tako doctor

tako doctor

Prints local diagnostics for the dev daemon, local DNS, TLS files, and platform-specific proxy setup. Global --json emits one object with paths, ca, dev_server, apps, and platform proxy checks instead of the human report.

tako run

tako run scripts/foo.ts
tako run jobs/foo.go --dry
tako run --eval 'console.log(tako.env)' -- --dry
tako run --env staging scripts/foo.ts
tako run -- cargo run --bin migrate

Runs a one-off command locally from the app directory with Tako project context. --env defaults to development.

Script files are expanded by the selected runtime. Bun runs JS/TS scripts with Bun, Node runs TS scripts with type stripping, and Go runs .go files with go run. --eval runs inline source when the runtime supports it; JS runtimes support inline TypeScript. Use -- {command...} for exact commands and runtimes without a bare script rule.

The child process receives [vars] plus [vars.<env>], ENV, TAKO_BUILD=local, TAKO_DATA_DIR, runtime defaults, and TAKO_APP_ROOT for JS apps. Tako decrypts local app secrets for the selected environment and passes the normal bootstrap envelope through TAKO_BOOTSTRAP_DATA. SDK-aware scripts use the same app SDK surfaces; JS/TS scripts can import tako and use tako.secrets and tako.storages.

Secrets are not process env vars. tako run is local-only in v0; it does not run commands on deployed servers.

Deploy And Runtime Operations

tako deploy [--env <env>] [-y|--yes] [--force] [--description <text>]

Builds locally, uploads artifacts over signed HTTP management, prepares every mapped server, runs the optional release command on the leader, rolls instances, and finalizes the release. Configured SQLite databases are backed up continuously; deploys do not create full-directory archives. --env defaults to production. development is reserved for tako dev. Deploy requires the CLI, server, and release artifact to use the same protocol version; a mismatch fails before release commands, workflows, or app processes start. --force attempts the deploy despite a mismatch but does not skip normal validation or readiness checks. It is separate from --yes, which only skips confirmation.

tako logs [--env <env>] [--tail] [--days N]

Shows app logs from all servers mapped to the environment. History defaults to three days. --tail streams. Global --json emits a final object with a logs array in history mode; with --tail, it emits JSONL records.

tako scale <instances> [--service <service>] [--env <env>] [--server <server>] [--app <app>]
tako scale --reset [--service <service>] [--env <env>] [--server <server>] [--app <app>]

Changes desired instance count on targeted servers. From a project, app identity is resolved from the config. Outside a project, pass --app, either as a base app with --env or as a full {app}/{env} deployment id.

For a Compose group with exactly one x-tako app, that app is the default scale target. With multiple apps, supply --service. Untagged Docker services can also be selected with --service. The count is a persistent override that survives deploys, rollbacks, and restarts. --reset clears the selected service override and applies its current Compose deploy.replicas default (one when omitted). Supply either a count or --reset. These flags require a Compose app. Scaling a stopped app or group updates its count for the next start without resuming it. See Compose scaling for limits and shared-volume behavior.

tako stop [--service <service>] [--env <env>] [--server <server>] [--app <app>]
tako start [--service <service>] [--env <env>] [--server <server>] [--app <app>]

Stops or starts an existing deployment. From a project, pass --env to use its configured servers, or select --server (defaulting to the production environment when --env is omitted). Outside a project, pass --server and either --app <app>/<env> or --app <app> --env <env>. Both commands support global --dry-run and --json.

For Compose, --service selects an x-tako app; omitting it selects the whole group, including its ordinary Docker services. Stopped apps retain their routes and configured replica counts. The stopped state survives deploys, rollbacks, and server restarts; requests do not wake them. Starting a group preserves apps that were individually stopped. Start the group before starting one of its apps. Explicitly stopped dependencies stay stopped even when a consumer starts or is redeployed. See Compose start and stop.

For native apps, stop also stops workflow workers. Unlike scale 0, it prevents automatic cold starts. Start resumes the configured instance count and workflows; for an app configured with zero minimum instances, it warms one instance without changing that minimum.

tako delete [--env <env>] [--server <server>] [-y|--yes]
tako rm ...
tako remove ...
tako undeploy ...
tako destroy ...

Deletes one deployment target, drains and stops the app, removes runtime state, and removes the server-side app data tree for that app/environment on the selected server.

A Compose service cannot be deleted independently: tako delete --service <service> is rejected before contacting a server. Remove the service from the Compose file and redeploy, or delete the whole group. Docker named volumes remain after group deletion.

Releases

tako releases list [--env <env>]
tako releases ls [--env <env>]
tako releases rollback <release-id> [--env <env>] [-y|--yes]

Release descriptions default to the commit subject for clean checkouts, Uncommitted changes based on: <commit subject> for dirty checkouts, or Local source snapshot without Git history. Set tako deploy --description "Fix reconnect handling" to override the description. The dirty status remains visible with a custom description. releases list --json includes description, commit_message, git_dirty, and sdk for each release. sdk is { "language", "name", "version" } for the Tako SDK detected at deploy time, or null when the release has none.

Lists release/build history or rolls an environment back to a previous release. Rollback uses the standard rolling-update flow and reports partial server failures individually.

Networks

tako network create <name>
tako network client add [--network <id-or-name>] [--admin | --deploy <app/env>] [--ssh] [--expires-in <seconds>]
tako network client list [--network <id-or-name>]
tako network client remove <device> [--network <id-or-name>]
tako network join [--name <device>] [--invitation-stdin]
tako network sync [network]
tako network request [--name <device>] [--network <id-or-name>] --output <file>
tako network approve <request-file> [--network <id-or-name>] --output <file> [--deploy <app/env>] [--admin] [--ssh] [--expires-in <seconds>]
tako network import <profile-file>
tako network export [network] --output <file>
tako network list --json
tako network status [network] --json
tako network revoke <device> [--network <id-or-name>]
tako network server drain <server> [--network <id-or-name>] [--timeout <seconds>]
tako network server resume <server> [--network <id-or-name>]
tako connect [network]
tako disconnect [network]

Invitations are single-use, expire after 3600 seconds by default (client add --expires-in accepts 60–86400), and are redeemed through their issuing server. join prompts for the invitation or reads stdin. network sync pulls published ciphertext without publishing worktree drafts. Network names are shared and currently immutable.

Each device generates and retains its own keys. Repeat --deploy for exact app/environment scopes; --ssh requires --admin. network approve --expires-in accepts 1..31536000 seconds and limits membership lifetime, separately from the request’s activation window. Import requires the original requesting device’s keys. Export contains public network configuration, not private keys or app secrets.

drain and resume require server administration and a unique member name or device ID. Drain defaults to 120 seconds and accepts 0..3600; zero requests immediate force. Inspect the structured completion report, not only the drained intent. See Native server maintenance.

ssh uses embedded WireGuard and userspace TCP without sudo or system network changes; OpenSSH is required. connect enables local applications, private DNS, short server names, and approved SSH aliases such as ssh edge-a. It explains the required administrator access before prompting. disconnect removes those managed settings. No external WireGuard executable is required. Normal native management uses pinned HTTPS without a tunnel or SSH agent. See Native networks for platform requirements, enrollment, endpoints, and private service names.

Server administrator recovery

Run on a surviving Linux server through its provider console or administrator SSH:

sudo tako-server admin invite [--expires-in <seconds>] [--ssh] [--json]

Requires root and a running native network agent with intact state and keys. Prints a single-use invitation for an independent administrator, with a default validity of 3600 seconds (60–86400 accepted). Join from the replacement machine with tako network join. --ssh also grants a Tako administrator shell. Ordinary client invitations remain delegated. See Recover administrator access.

Servers

tako network server add [admin@host] [--network <id-or-name>] [--name <name>] [--wireguard-endpoint <ip:udp-port>] [--port <ssh-port>] [--ssh-key <path>] [--http-port <port>] [--https-port <port>]
tako network server list
tako network server ls
tako status
tako network server reload <name> [--force]
tako network server upgrade [name] [--force]
tako network server remove [name]
tako network server rm [name]
tako network server uninstall [name] [-y|--yes]

network server add bootstraps over administrator SSH and enrolls the server’s own keys, HTTPS pins, and endpoint in a Tako network. No [[servers]] entry is written. Without --network, selection uses the project’s production network (honoring --config) or the sole configured network. An explicit --network takes precedence. Create or import a network first; ambiguous selection requires --network.

Without a host, an interactive terminal prompts for admin@host. An omitted SSH user defaults to root. Public HTTP and HTTPS ports default to 80 and 443. --ssh-key selects a private key for bootstrap and saves its absolute path locally for SSH maintenance (reload, upgrade, uninstall). This selection stays on the workstation; normal HTTPS management uses the enrolled device identity. External VPN providers and the old --vpn, --install, --no-test, and server --description options are removed. See Servers to replace an old setup.

status prints a deployment snapshot grouped by server, with compact server summary, routes, and app rows. An app whose instances the server had to replace shows a Restarts row with the restart count, the last exit (crashed, unhealthy, or failed to start, with its exit code or signal), and when the next restart is due while restarts are backing off. --json reports the same under recovery in each app status.

Use the exact selector shown by network server list --json for maintenance commands. Usually network-name/server-name, it includes network or server IDs when names collide; see Server maintenance. network server reload performs a zero-downtime service reload by default. Its --force flag performs a full restart. network server upgrade checks the candidate protocol against the CLI and every active app before reload, checks again after readiness, and restores and restarts the previous binary if the upgrade fails. network server upgrade --force attempts an incompatible upgrade but keeps readiness and rollback checks enabled. network server uninstall publishes a revocation of native membership and delegated approvals before removing the remote service, binaries, data, and sockets. Revocation failure stops uninstall before deletion; cleanup failure after revocation reports uninstall as incomplete. Offline peers enforce the revocation when they reconnect. --dry-run performs neither revocation nor deletion. network server remove revokes network membership and delegated approvals without deleting app data.

Secrets

tako secrets set <NAME> [--env <env>] [--expires-on <when>]
tako secrets add <NAME> ...
tako secrets rm <NAME> [--env <env>] [--yes]
tako secrets list
tako secrets sync [--env <env>] [--retry-expired]
tako secrets migrate
tako secrets key export [--env <env>]
tako secrets key import [--env <env>] [--passphrase]

Secrets are app-facing values delivered to SDKs. Values are encrypted in ignored worktree-local .tako/secrets.local.json; keys are stored outside the project. Interactive prompts accept pasted multiline values and ask before trimming surrounding whitespace; non-interactive commands read the full stdin stream. Set/remove commands only edit local drafts. Deploy and secrets sync pull published state, preserve local edits, and publish per-key changes for the selected environment. Sync does not restart running apps; deploy to apply values. Network-bound secrets, provider credentials, storage credentials, and backup keys are replicated as ciphertext between servers. Authorized clients cache them through HTTPS.

Secret state retains the current value per key and versioned deletions. Pending publication requests expire after seven days; --retry-expired explicitly replaces expired requests with the current local drafts as new changes. Review the drafts first, since they can overwrite remote changes to those keys.

secrets migrate validates the original environment keys, writes local state, and removes the old .tako/secrets.json after retaining an ignored recovery copy. Commit the tracked file’s deletion. Historical commits remain unchanged. See Secrets for conflicts, worktrees, and recovery.

secrets key import can recover the original environment key before local drafts are merged with cached shared values. It accepts an exported key through a masked prompt or stdin; use --passphrase --env <env> to derive the original key from its passphrase instead. It rejects keys that cannot decrypt existing values. An invitation restores access to published values, but does not recover unpublished drafts or legacy environment keys.

Expiry accepts YYYY-MM-DD or in N days. Leave it blank for no expiry. Deploy fails on expired selected-environment secrets and warns when they expire within 30 days.

Provider Credentials

tako credentials set [name] [--env <env>] [--expires-on <when>]
tako creds set ...
tako credentials rm <name> [--env <env>]
tako credentials list

Running tako credentials without a subcommand lists credentials. Set and remove prompt for an environment when --env is omitted; non-interactive use requires --env.

Provider credentials are Tako-owned runtime credentials, not app secrets. Supported names are ssl.cloudflare and postgres_url. They are encrypted in local state and shared network records, never exposed to app code, and published by tako secrets sync or deploy. Running processes receive the new bindings on deploy.

ssl.cloudflare powers Let’s Encrypt DNS-01 and Cloudflare Origin CA. postgres_url enables shared channel replay, Presence membership, Store invalidation, and workflow state for multi-server deployments. It is a provider credential and is not exposed to app code.

Storage

tako storages add <binding> [--env <env>] [--resource <resource>] [--provider s3|local] [--bucket <bucket>] [--endpoint <https-url>] [--region <region>] [--access-key-id <value>] [--secret-access-key <value>] [--expires-on <when>] [--force-path-style] [--public-base-url <https-url>]
tako storages credentials <resource> [--env <env>] [--access-key-id <value>] [--secret-access-key <value>] [--expires-on <when>]

Storage commands default to --env production and --provider s3.

storages add writes the app binding to tako.toml. For S3 resources it also writes resource metadata and encrypted credentials. storages credentials rotates credentials for an existing top-level S3 resource without adding an app binding, which is useful for backup-only resources.

Backups

tako backups now [--env <env>] [--server <server>]
tako backups list [--database <path> [--shared]] [--env <env>] [--server <server>]
tako backups ls [--env <env>] [--server <server>]
tako backups status [--env <env>]
tako backups download <backup-id> [--env <env>] [--server <server>] [--output <path>]
tako backups restore <backup-id> [--env <env>] [--server <server>] [-y|--yes]
tako backups restore --database <path> (--latest|--txid <capture-id>) [--shared] [--env <env>] [--server <server>] [-y|--yes]

Backup commands default to production, require project context, and target the deployment id {app}/{env} over signed HTTP management. Multi-server download and restore require --server.

backups now creates a manual full app-directory archive; deploys and timers do not create one. backups list lists these archives and older databases archives. Add --database <path> to list continuous SQLite recovery points and --shared for paths relative to TAKO_SHARED_DATA_DIR. backups status shows capture/upload progress, errors, and consuming apps for each selected database, plus manual archive and unfinished-restore status.

Use backups restore --database <path> --latest or --txid <capture-id> to restore SQLite. Add --shared for a shared database; this requires server-admin permission and pauses all apps on that server. Private database restores pause the owning app. Use --server when multiple servers are mapped. backups restore <backup-id> restores a manual archive, including workflow state when present. download saves an encrypted archive; continuous SQLite history uses the database restore commands. See Backups for retention, restore locks, and recovery on a replacement server.

Workflows

tako workflows list [--env <env>] [--server <server>] [--status <status>] [--name <workflow>] [--limit <n>]
tako workflows ls ...
tako workflows show <run-id> [--env <env>] [--server <server>]
tako workflows retry <run-id> [--env <env>] [--server <server>]
tako workflows cancel <run-id> [--env <env>] [--server <server>] [-y|--yes]

Workflow commands inspect and repair the durable runs of a deployed environment. They default to production, require project context, and target the deployment id {app}/{env} on every mapped server, or only on --server. They are not available for development; use the tako dev logs locally.

list shows the newest runs first: ID, workflow, status, attempts used of the retry budget, the next due time of a pending run or the finish time of a finished one, and the first line of the last error. --status accepts pending, running, succeeded, cancelled, or dead. --name selects one workflow. --limit accepts 1..200 and defaults to 50. show prints every run field, the payload and result, up to the 100 most recent saved steps oldest first, event waits, and the stored error. Servers cap large values; a capped value shows the start of its JSON text and its stored size.

retry accepts a dead or cancelled run. The run becomes pending, due now, with a fresh retry budget; saved steps are kept, so it resumes after its last checkpoint. cancel accepts a pending or running run and asks for confirmation in an interactive terminal unless --yes is passed. A running handler is not interrupted: it stops at its next checkpoint (the next step, sleep, or wait) and cannot complete the run; side effects already started are not undone. Retry does not prompt. Both act on exactly one server.

Without postgres_url, each server keeps its own runs, so results are labelled with their server; if two servers hold a run with the same ID, show, retry, and cancel fail and ask for --server. With postgres_url, every server shares one store: runs are listed once, without a server, and retry or cancel is sent once. A server that cannot be reached is reported as a warning while the others answer; the command fails only when no server answers, or when the run is not found on any server.

Global --json prints one object. Every object has ok, command: "workflows", app, environment, store ("sqlite" or "postgres"), and errors, an array of {name, error} for servers that did not answer.

  • list adds runs and has_more. Each run has id, name, status, attempts, max_attempts, created_at_ms, run_at_ms (null while waiting for a signal), finished_at_ms, last_error (capped at 1 KiB), last_error_truncated, and server (null with Postgres).
  • show adds server and run. run has the list fields plus lease_until_ms, worker_id, payload, result (null until the run succeeds), steps (name, completed_at_ms, result), steps_total, and waits (step_name, event_name, expires_at_ms). Each value is {value, truncated, bytes, preview?}; a truncated value has value: null and a preview of its JSON text.
  • retry and cancel add server, id, previous_status, and status.

Failures print {"ok": false, "error": {"message": ...}}, as for every command.

Local CLI Maintenance

tako completions [-y|--yes]
tako upgrade
tako uninstall [-y|--yes]
tako version

completions installs bash, zsh, and fish completion scripts for shells that are on PATH. zsh is installed only when a writable directory is already on fpath; otherwise Tako prints how to add one. upgrade updates the local CLI install. On macOS, official CLI upgrades support Apple Silicon and Intel. Intel builds publish separately after each release. If an Intel download is temporarily unavailable, retry once that build finishes. uninstall removes local Tako binaries, local data, completion scripts, and platform-specific dev services/config after confirmation.

Usage stats

Official CLI installs send one anonymous event per command so we can count unique users and see which commands run (deploy, dev, init, and so on). The event includes CLI version, OS, architecture, and the command name. It does not include app names, paths, server hosts, environment names, or command arguments. tako-server does not send usage stats.

Opt out:

export TAKO_TELEMETRY=0

Usage stats are always off during dry runs and in CI (CI=true or --ci), even with TAKO_TELEMETRY=1. Local Cargo builds are off by default; set TAKO_TELEMETRY=1 to send from local builds outside CI.

Last updated on

Was this page helpful?