⏳ This skill is pending AI review.
Scores will appear once the review pipeline completes.
d1v
Operate d1v.ai projects and container workspaces with the d1v CLI. Use when a task involves D1V authentication, project binding, workspace synchronization, container commands, environment metadata, preview deployments, deployment status, or a user-confirmed production release.
Choose how to use this skill
You do not need every option. Choose the path your AI client supports. The stable page stays the same; versioned files are immutable.
1. Native installer
This listing has no registered native installer command. Use the complete package or source fallback below, depending on what your client supports.
Do not guess an installer command or replace an existing version without reviewing the diff.
2. Complete package recommended
Download the ZIP when available. It includes SKILL.md plus the references, security notes and version metadata.
No complete ProSkills package is published for this listing yet.3. Prompt-only
Copy the prompt above when the agent can read the stable page or when you want to adopt the workflow without installing a skill.
Need only the instruction file?
Download SKILL.md only if your client requires a single file. The complete ZIP is safer for a full installation because it preserves the references and release context.
No path installs or executes anything by itself. Your agent still needs access to the project files. Before updating, compare the installed version and review the diff.
// RATINGS
// README
D1V CLI
Deploy, inspect, and work inside D1V projects from your terminal.
Project-aware by default. Works with Codex and Claude Code. Supports cloud and local runtimes.
Quick Start
Install, authenticate, and deploy the current directory:
curl -fsSL https://d1v.ai/install/d1v-cli.sh | bash
d1v auth login --browser
d1v --preview # or: d1v --prev
d1v --prod
The CLI reads D1V_PROJECT_ID from the current directory's .env first, then
from the process environment and .d1v/project.json. Explicit project
arguments always win. When no project can be resolved interactively, choose one
from the project picker. Cloud environment variables are merged into .env;
local values are kept by default when keys conflict.
Run d1v --help for the complete command list.
Coding Agent Skill
The official, versioned d1v Skill lives at
skills/d1v/SKILL.md. It gives Codex and Claude Code
safe instructions for project workspaces, container commands, preview
deployments, and user-confirmed production releases.
The curl installer uses --install-skill auto by default. It installs only
for Codex and Claude Code executables already available on PATH; it does not
install either coding agent and succeeds without changes when neither is found.
Use the same behavior manually:
d1v skill install --agent auto
Use --agent codex, --agent claude, or --agent all to explicitly choose
targets. Skills are written below ${CODEX_HOME:-~/.codex}/skills/d1v and
${CLAUDE_CONFIG_DIR:-~/.claude}/skills/d1v. An identical Skill is left alone;
a different existing SKILL.md is backed up beside it as
SKILL.md.d1v-backup-<UTC timestamp> before replacement. The legacy
https://www.d1v.ai/cli-skill.md URL remains available and redirects to this
canonical file.
D1V Mobile App
Pair the terminal workflow with d1vai_app,
the official open-source Flutter client for d1v.ai. Use the CLI for fast local
and CI workflows, and the mobile app to create projects, continue AI sessions,
inspect files, monitor deployments, and manage your workspace away from a desk.
The companion app supports iOS and Android and includes its own screenshots, release downloads, and developer guide:
Local Runtime
D1V supports a local runtime in addition to the existing cloud runtime.
Role split:
opcode-api: the local runtime serverd1v-cli: installer, launcher, supervisor, connectorbackend_admin: control plane and runtime routerd1vai: unified frontend
The frontend never connects directly to the user machine. The local agent opens an outbound connection to D1V cloud.
Local Runtime Quickstart
- Install and inspect the local runtime:
d1v runtime doctor
bash scripts/install-opcode-runtime.sh --home ~/d1v-home
- Pair the machine:
d1v agent pair
Or keep using a pairing code generated in the web UI:
d1v agent pair --code <pairing-code>
- Start the local runtime:
d1v agent start
- Create or bind local project directories:
d1v agent project create --project-id <project_id> --name my-app
d1v agent project import --path ~/work/my-app --project-id <project_id>
d1v agent project bind --project-id <project_id> --path ~/work/my-app
Backward-compatible entry:
d1v agent init-runtime --project-id <project_id> --path ~/work/my-app
Local Runtime Behavior
- Runtime switching affects new sessions only.
- Existing sessions stay pinned to the runtime where they started.
- If a project is bound to local runtime and the device is offline, requests return an explicit local-runtime error instead of silently falling back to cloud.
- Project creation/import flows may still use cloud opcode directly before runtime binding exists.
Public Expose
CLI free expose does not require a running node agent:
d1v expose 3000
d1v expose list
d1v expose close <binding_id>
Platform node ingress stays on a separate command path:
d1v node expose 3000 --node-id <platform-node-id>
d1v node expose list
d1v node expose close <binding_id>
Current expose modes:
cli_free_relay: returns a publichttps://*.cli-free.d1v.devURL for login-backed temporary CLI relayscloudflare_tunnel: returns a publichttps://*.node.d1v.devURL for platform nodesreverse_relay: internal fallback mode used by local/customer relays behind the CLI-free entry
d1v expose currently targets HTTP traffic. Browser terminal and session WebSocket flows still use the existing backend relay path.
Platform Node
Platform nodes now use a single control origin for runtime-agent ingress. The recommended bootstrap form is:
d1v node start --key <platform-node-key> --control-origin https://{your-host-or-name}-node.d1v.dev
If --control-origin is omitted, the runtime-agent falls back to its own public IP detection and control-plane registration flow.
Privacy Boundary
Current phase:
- paired devices only
- outbound-only relay
- short-lived pairing code
- stored device public key
Current limitation:
- relay traffic is not zero-knowledge
- platform metadata and chat/session records still follow existing D1V persistence behavior
For deeper architecture details, see docs/d1v-agent-architecture.md.
Install
Recommended:
curl -fsSL https://d1v.ai/install/d1v-cli.sh | bash
Install page:
https://d1v.ai/cli-install
Alternatives:
brew install d1vai/tap/d1v
cargo binstall d1v-cli
cargo install --locked d1v-cli
After install:
d1v auth login
d1v project list
d1v github status
To authenticate through an existing browser session and save a revocable device
API key, use d1v auth login --browser. The browser approves a one-time,
10-minute session; the CLI stores the resulting key in ~/.d1v/config.toml.
Upgrade later:
d1v upgrade
d1v upgrade --version v0.1.5
d1v uninstall
d1v upgrade replaces only the executable. Your login remains in
~/.d1v/config.toml; upgrading does not sign you out or copy credentials into
the release archive.
Global Options
| Option | Description | Default |
|---|---|---|
--format | Output format (text, json) | text |
--color | Color output | auto |
--lang | Display language | System / Config |
-v | Increase log verbosity | warn |
Environment Variables
| Var
// HOW IT'S BUILT
// FILES
2 files shipped with this skill. Inline text is the catalog copy. The upstream link is the source repository.
skills/d1v/SKILL.md5.0 KBd7759c406684upstream ↗
--- name: d1v description: Operate d1v.ai projects and container workspaces with the d1v CLI. Use when a task involves D1V authentication, project binding, workspace synchronization, container commands, environment metadata, preview deployments, deployment status, or a user-confirmed production release. --- # d1v CLI Use the installed `d1v` command. For automation and structured inspection, use `--format json`; never mix JSON output with an interactive shell. ## Safety - Never print, log, commit, or add API keys, tokens, environment values, or shell tickets to source files, command output, or repository configuration. - Inspect authentication and project state before changing anything. Check the command exit code and query resulting state; a submitted request is not proof of success. - Run `pull --dry-run` or `push --dry-run` before changing a local workspace, then inspect repository instructions and local changes. - Treat preview and production as separate environments. A production release requires an interactive TTY and explicit user confirmation; do not attempt or imply a non-interactive production release. - If a deployment fails or times out, preserve its deployment/release ID and use `deploy status` or `deploy history` rather than blindly retrying. ## Authenticate and discover ```bash d1v auth status --format json d1v project list --format json d1v github status --format json ``` If authentication is missing, ask the user to run `d1v auth login`. Do not ask them to paste a token into chat or a command line. An API key supplied through a secure standard-input workflow can be used with `d1v auth login --api-key`. ## Bind a local workspace ```bash d1v project get <project_id> --format json d1v init . --project-id <project_id> d1v pull --dry-run --format json d1v pull d1v push --dry-run --format json d1v push ``` `d1v init` writes local workspace metadata. Do not use `--force` unless the user explicitly approves replacing an existing binding. ## Run commands in a workspace Use an interactive shell only when a person needs a terminal: ```bash d1v shell d1v shell <project_id> d1v shell --organization-id <organization_id> ``` - With no target, `d1v shell` opens the personal workspace root. - A project ID opens that project's directory. Do not combine it with `--organization-id`. - Interactive shells require a TTY and text output. Do not use `--format json` with `d1v shell`. For automation, Agent actions, CI, or captured exit status, use `d1v exec` and place the remote argv after `--`: ```bash d1v exec -- git status --short d1v exec --project-id <project_id> -- npm test d1v exec --organization-id <organization_id> -- pwd d1v --format json exec --project-id <project_id> -- sh -c 'printf ok; printf problem >&2; exit 7' ``` Read `stdout`, `stderr`, `cwd`, and `exit_code`; a non-zero remote exit is a failed action. Do not wrap commands in `sh -c` unless shell syntax is required. Shell access uses a short-lived, single-use ticket. Never print, persist, forward, or reconstruct terminal WebSocket URLs from that ticket. ## Environment metadata ```bash d1v env list --project <project_id> --format json d1v env set --project <project_id> KEY=VALUE --sensitive ``` The default list output keeps sensitive values redacted. Do not use `--reveal`, `env get --reveal`, or `env export` unless the user explicitly asks for the secret material and has provided a safe destination outside version control. ## Deployments Create a preview with the explicit command when a project ID is known: ```bash d1v deploy preview <project_id> --format json d1v deploy status <project_id> --format json d1v deploy history <project_id> --environment preview --format json ``` `d1v deploy preview` waits for the deployment's terminal state. Report success only when it returns READY, and use its `production_url` when present, otherwise its `vercel_url`. On failure, report the error and deployment ID without repeating the request automatically. `d1v --preview` (or `d1v --prev`) is the current-directory shortcut. It may prompt to select or create a project and write the resulting project binding; use it only when the user wants that local-directory workflow. For production, first explain the pending operation and then require a clear user confirmation in an interactive terminal: ```bash d1v deploy prod <project_id> ``` The CLI shows release preflight information, prompts for relevant environment decisions, creates an idempotent release, and waits for its terminal state. It rejects non-interactive production releases. On failure, report the release phase, error code, and message. The `d1v --prod` shortcut follows the same confirmation and release flow for the bound current directory. ## Local runtime ```bash d1v runtime doctor --format json d1v agent status --format json d1v agent pair d1v agent start ``` Do not silently switch a requested local runtime to cloud when it is unavailable. Explain the runtime status and ask the user how to proceed. ## References - https://github.com/d1vai/d1v-cli - https://www.d1v.ai/cli-install - https://www.d1v.ai/docs/cli - https://www.d1v.ai/openapiREADME.mdupstream ↗
Contents are not inlined in the catalog. Open the upstream file.
Plain-text copy, no login: /skill-bundles/d1vai-d1v-cli-skills-d1v.md