@mcpcloud/cli

The official command-line interface for MCPCloud — build, deploy, and operate MCP servers and Skills from your terminal. Everything the dashboard does, scriptable: ingest an OpenAPI/GraphQL spec into a typed MCP server, deploy it to the edge, watch metrics, manage custom domains, and wire skills into your coding agent.

npm install -g @mcpcloud/cli
mcp login          # opens the browser sign-in — works out of the box
mcp whoami         # confirm your identity, org, and active profile

The package installs three equivalent binaries — mcp (primary), mcpsh, and mcpcloud. They all point at the same executable; use whichever you prefer.

Requirements

• Node.js 18 or newer
• An MCPCloud account (sign up at https://mcpcloud.sh)

Installation

# global install
npm install -g @mcpcloud/cli
bun  add     -g @mcpcloud/cli

# one-off invocation (no install)
npx  -y      @mcpcloud/cli whoami
bunx --bun   @mcpcloud/cli whoami

Quick start

The CLI ships with the MCPCloud production base URL baked in, so a fresh machine works with zero configuration — just sign in:

mcp login                       # browser sign-in (recommended)
mcp login --key mck_xxx         # non-interactive (writes to shell history — prefer the env var)
export MCPCLOUD_API_KEY=mck_xxx # or use an env var for CI

mcp whoami                      # identity, resolved base URL (marked "(production)"), active profile
mcp servers list                # your deployed MCP servers
mcp servers get srv_123         # detail view
mcp tools list --server srv_123 # tools that server exposes
mcp skills connect skl_123 --agent claude-code --apply
                                #   auto-register a skill with Claude Code

Run mcp <command> --help for the full option list, or mcp help <topic> for guided walkthroughs.

Pointing at staging or a self-host

The production default is only a default — override it per environment. Precedence is --base-url flag → MCPCLOUD_BASE_URL → the saved profile's baseUrl → the built-in production default:

export MCPCLOUD_BASE_URL=https://staging.example.com   # env (CI / scripts)
mcp config set-url https://staging.example.com         # persist to the active profile
mcp --base-url https://staging.example.com whoami      # one-off override

Authentication & config

Credentials and settings live in ~/.mcpcloud/config.json (mode 0600; respects XDG_CONFIG_HOME on Linux):

mcp login                       # save an API key (interactive browser flow)
mcp logout                      # remove the saved key
mcp config show                 # print the resolved config (key redacted)
mcp config current              # alias of `config show`
mcp config set-org <name|id>    # persist a default organization (accepts name, slug, or id)

MCPCLOUD_API_KEY overrides the saved key; MCPCLOUD_ORG_ID overrides the saved default org.

Profiles

Keep separate credentials + base URLs for prod, staging, and self-host, and always know which one you're about to mutate:

mcp config profile add staging --url https://staging.example.com --key mck_stg
mcp config profile list         # ✓ marks the active profile
mcp config use staging          # switch the active profile
mcp --profile prod servers list # one-off override
export MCPCLOUD_PROFILE=staging # or select via env

mcp whoami prints the active profile and its source, and appends (production) to the base URL when it resolves to the production host — so a profile named "local" pointed at prod can't surprise you. mcp doctor runs the same resolution as an 8-point health check.

The dev loop

mcp dev runs a deployed server locally with hot reload: it downloads the server's generated bundle, serves it on localhost (HTTP/SSE/stdio), watches your files (and optionally an OpenAPI spec) to regenerate on change, auto-wires the running instance into your coding agent (--auto-connect claude-code), and exposes an inspector UI at /__inspect. Pair it with the round-trip tool editors (tools pull / tools edit / tools diff / tools reset-handler) to iterate on tool metadata and handler source against the live bundle.

Scripting & CI

Every command honors a machine-readable output contract, registered globally so it's uniform across the tree.

Output formats

| Flag | Effect | | -------------------- | ----------------------------------------------------------------------------------- | | --json | Emit a single JSON document on success; errors become a { "error": {…} } envelope. No human text is mixed in. | | --format <mode> | List output as table (default), tsv, jsonl, or json. | | --no-table | Shorthand for --format tsv — tab-separated rows for awk/cut/grep. | | --field <names> | Comma-separated columns to include in list output (repeatable). | | --filter <k=substr>| Case-insensitive substring filter on list rows (repeatable). | | --quiet | Suppress success/step/info chrome; data and errors still print. Pairs with exit codes for gates. | | --non-interactive | Refuse prompts (fail fast on missing input) without the color/glyph changes of --ci. | | --ci | Force CI mode (auto-detected when CI=true): no color, ASCII glyphs, no prompts, always surface request IDs. | | --warnings-as-errors | Exit non-zero if any warning is emitted. |

mcp --json servers list | jq '.servers[].id'
mcp servers list --field id,name,status --format jsonl
mcp servers list --filter status=active --no-table | cut -f1

Exit codes

| Code | Meaning | | ---- | ---------------------------------------------------------------------------------------- | | 0 | Success. | | 1 | Generic failure (API error, validation error, network error). | | 2 | No API key configured. Run mcp login or set MCPCLOUD_API_KEY. | | 3 | No API base URL configured (only reachable if you clear the built-in default and set nothing). |

Error envelope

Structured API errors are surfaced consistently. In --json mode:

{ "error": { "code": "server_not_found", "message": "Server not found", "requestId": "api_018f…", "docsUrl": "https://mcpcloud.sh/docs/errors#server_not_found" } }

In human mode the same fields print as code, a one-line remediation, the request id (always shown under --ci, even when the upstream response omitted one), and an absolute docs: link.

Idempotency

Mutations accept --idempotency-key <key> (1–255 ASCII chars, no whitespace). When omitted, the CLI auto-generates a UUID per invocation so its own internal retries on transient failures stay safe. Pass a stable key from your pipeline to make a whole step re-runnable.

CI deploy gate

The observability commands compose into a deploy gate that exits non-zero on failure — no output parsing required:

mcp servers deploy srv_123 --wait                        # block until the deploy is terminal
mcp deployments health <deploymentId> --quiet            # exit non-zero unless healthy
mcp metrics deployment <deploymentId> --since 15m --max-error-rate 0.02
                                                         # exit non-zero if the error rate breaches the budget

Resolve the deployment id from mcp servers get srv_123 --json (or mcp deployments list) after the deploy completes.

Command reference

Run mcp <command> --help for full options on any of these.

Generated from the live command tree by bun run docs:readme — do not edit by hand.

Top-level commands

| Command | Description | | --- | --- | | mcp analytics | Show the workspace analytics rollup (readiness, funnel, usage) | | mcp completion <shell> | Print a shell completion script (bash, zsh, or fish) | | mcp connect-agent | Connect your agent to the official MCPCloud MCP server (headless twin of /connect-agent) | | mcp doctor | Diagnose the CLI environment: Node, config, base URL, API key, claude CLI, and updates | | mcp help [topic] | Show extended help for a topic (auth, profiles, agents, errors) | | mcp init | Guided onboarding: pick org → pick/create project → create a server (empty skeleton, or generated from an OpenAPI/GraphQL spec via --from-spec) | | mcp invoke <server> [tool] | Call a tool on a deployed MCP server — or, with --local, boot this directory’s bundle once and call it there. | | mcp login | Sign in via browser (default) or paste an API key | | mcp logout | Remove the saved API key from ~/.mcpcloud/config.json | | mcp pull | Materialize a server’s working tree here: every tool as .md, every handler as .ts, plus the sync baseline push diffs against | | mcp push | Push local tool metadata and handler edits back to the cloud (diffed against your last pull) | | mcp ui | Interactive terminal UI: browse orgs/servers/skills, tail deployment logs, and watch the live mcp dev inspector | | mcp update | Check for a newer @mcpcloud/cli release and re-install it via your package manager | | mcp whoami | Show the caller identity, API key, base URL, and organizations you can access |

mcp alias — Manage per-profile resource shortcuts (aliases)

| Command | Description | | --- | --- | | mcp alias list | List aliases in the active profile | | mcp alias rm <name> | Remove an alias from the active profile | | mcp alias set <name> <target> | Create or update an alias. Target is :<id|name|slug>. |

mcp api-keys — Manage personal API keys

| Command | Description | | --- | --- | | mcp api-keys create | Create a new API key (the secret is shown once) | | mcp api-keys describe <apiKey> | Show one API key (accepts id or name; status, expiry, last used) — never the secret | | mcp api-keys list | List active API keys for the current user | | mcp api-keys revoke <apiKey> | Permanently revoke an API key (accepts id or name) | | mcp api-keys rotate <apiKey> | Mint a replacement key and grace-expire the old one (24h window; accepts id or name). The new secret is shown once. |

mcp audit — Workspace audit log (admin/owner only)

| Command | Description | | --- | --- | | mcp audit list | List recent audit events for an organization |

mcp config — Manage CLI configuration in ~/.mcpcloud/config.json

| Command | Description | | --- | --- | | mcp config clear-app-url | Forget the saved dashboard URL (falls back to MCPCLOUD_APP_URL or auto-derive) | | mcp config clear-editor | Forget the editor open preference, the picked editor command, AND the saved target (mcp dev will re-detect and prompt next session) | | mcp config clear-org | Remove the saved default organization ID | | mcp config clear-url | Remove the saved base URL (falls back to MCPCLOUD_BASE_URL) | | mcp config current | Alias for config show — print the active profile | | mcp config profile | Manage named config profiles (apiKey + baseUrl + defaultOrganizationId) | | mcp config profile add <name> | Create a new profile (use --url / --key / --app / --org to seed values) | | mcp config profile list | List all configured profiles | | mcp config profile remove <name> | Delete a profile (cannot remove the active or only-remaining profile) | | mcp config set-app-url <url> | Save the dashboard origin used by mcp login browser flow | | mcp config set-editor <pref> | Save the dev-runner editor open preference: always | never | | mcp config set-editor-command <command> | Override the editor launch command (default: code; e.g. cursor, code-insiders) | | mcp config set-org <organization> | Save a default organization (by name, slug, or id) | | mcp config set-url <url> | Save the API base URL to ~/.mcpcloud/config.json | | mcp config show | Print the active profile (API key is redacted) | | mcp config use <name> | Switch the active profile (e.g. staging, prod) |

mcp connections — Manage your per-server upstream API keys (BYOK) from the terminal

| Command | Description | | --- | --- | | mcp connections add <server> | Save (or replace) your upstream API key for a server — rotation is just adding again | | mcp connections list | List your stored upstream keys across servers (hints only) | | mcp connections remove <server> | Revoke your stored upstream key for a server |

mcp deployments — Inspect deployments

| Command | Description | | --- | --- | | mcp deployments get <deployment> | Show one deployment by id or selector (latest | active | previous, with --server) | | mcp deployments health <deployment> | Show a deployment's last health-check result (status, HTTP code, when it was checked) | | mcp deployments list | List deployments in an organization | | mcp deployments logs <deployment> | Show recent deployment events; pass --follow to tail new ones as they arrive | | mcp deployments rollback <deployment> | Roll back to a prior deployment — flips a paused/undeployed deployment back to active and demotes the current latest in one shot. | | mcp deployments runtime-auth <deployment> | Inspect a deployment's runtime-auth posture (access mode, exchange path, edge policy) |

mcp dev — Run a deployed MCP server locally with hot reload + agent auto-wire

| Command | Description | | --- | --- | | mcp dev init | Interactively pick an organization + server and write .mcpcloud/state.json (forces the picker even if state already exists) | | mcp dev invoke <tool> | Call a tool against the running mcp dev session in this directory. No auth — local runtime accepts anonymous JSON-RPC. | | mcp dev kill <pid> | Stop a running mcp dev session by PID (or "all" to stop every session) | | mcp dev list | List running mcp dev sessions on this machine | | mcp dev replay | Replay one or more recorded calls against the running mcp dev server | | mcp dev tail | Stream new mcp dev inspector records to stdout as JSON lines (Ctrl+C to stop) |

mcp domains — Serve servers from custom domains you own (Pro+)

| Command | Description | | --- | --- | | mcp domains add <hostname> | Attach a custom domain to a server (admin role required) | | mcp domains list | List the custom domains in an organization | | mcp domains namespaces | Verify a parent domain once, then attach subdomains to any server with no further DNS work | | mcp domains namespaces add <parentDomain> | Register a domain namespace and print its one-time DNS records (admin role required) | | mcp domains namespaces list | List the domain namespaces in an organization | | mcp domains namespaces remove <parentDomain> | Remove a domain namespace — existing attached hostnames keep working (admin role required) | | mcp domains namespaces verify <parentDomain> | Check the namespace verification records now | | mcp domains remove <hostname> | Remove a custom domain — traffic on it stops immediately (admin role required) | | mcp domains status <hostname> | Recheck a custom domain against Cloudflare and show status | | mcp domains transfer <hostname> | Move a custom domain to another server without certificate churn (admin role required) |

mcp installation — Manage your skill / artifact installations

| Command | Description | | --- | --- | | mcp installation disable <installationId> | Disable an installation by id (replay-as-success on already-disabled rows) | | mcp installation list | List your active installations and their ids | | mcp installation remove <installationId> | Mark an installation as removed by id |

mcp marketplace — Browse the public registry of MCP servers and skills

| Command | Description | | --- | --- | | mcp marketplace fork <artifact> | Fork a registry artifact (id or name) into one of your projects as an editable draft | | mcp marketplace get <artifact> | Show full details for one registry artifact (accepts id or name; readme, recent versions) | | mcp marketplace install <artifact> | Install a published registry artifact (server or skill; accepts id or name) into your organization | | mcp marketplace list | List published artifacts in the registry | | mcp marketplace publish <server|skill> | Publish a deployed server or a versioned skill to the registry (requires org admin) | | mcp marketplace version <artifact> <versionId> | Show details for one published version of a registry artifact (artifact accepts id or name) | | mcp marketplace versions <artifact> | List published versions of a registry artifact (accepts id or name) |

mcp metrics — Aggregate deployment metrics (request volume, error rate)

| Command | Description | | --- | --- | | mcp metrics deployment <deployment> | Show aggregate request metrics for a deployment over a window (CI deploy gate) |

mcp oauth — Manage OAuth provider connections

| Command | Description | | --- | --- | | mcp oauth connection | Inspect or revoke individual OAuth connections | | mcp oauth connection delete <connection> | Revoke an OAuth connection by id or provider key (best-effort upstream revoke; replay-as-success on already-revoked rows) | | mcp oauth connections | List OAuth provider connections across the organization |

mcp orgs — List and inspect the organizations you belong to

| Command | Description | | --- | --- | | mcp orgs get [organizationId] | Show one organization (defaults to your active org) | | mcp orgs list | List organizations you can access | | mcp orgs members [organizationId] | List members of an organization | | mcp orgs usage [organizationId] | Show request-volume usage for an organization | | mcp orgs use <organizationId> | Set the default organization for this profile (discoverable alias of mcp config set-org) |

mcp plugin — Manage @mcpcloud/cli plugins (third-party command extensions)

| Command | Description | | --- | --- | | mcp plugin install <package> | Install a plugin from npm (uses npm under the hood) | | mcp plugin list | List installed plugins | | mcp plugin remove <name> | Remove an installed plugin |

mcp projects — Manage projects

| Command | Description | | --- | --- | | mcp projects api-source <apiSourceId> | Show details for a single imported API spec | | mcp projects api-sources <project> | List imported API specs / source material for a project | | mcp projects create | Create a new project in an organization | | mcp projects delete <project> | Permanently delete a project. Requires --confirm "" and that the project is empty of servers/skills. | | mcp projects get <project> | Get details for a single project (accepts id or name) | | mcp projects list | List projects in an organization | | mcp projects update <project> | Update a project (RFC 7396 merge patch — only listed fields change) |

mcp servers — Manage MCP servers

| Command | Description | | --- | --- | | mcp servers create | Create an empty server skeleton in a project | | mcp servers delete <server> | Permanently delete a server and all its data. Requires --confirm "". | | mcp servers deploy <server> | Deploy a server's generated bundle to Cloudflare Workers (closes the spec → deploy loop) | | mcp servers deployments <server> | List a server's deployment history | | mcp servers deps | Compose another server's tools into this one (list / add / rm) | | mcp servers deps add <server> <child> | Compose 's tools into . Both accept an id, name or slug. | | mcp servers deps list [server] | List the servers this one composes. | | mcp servers deps rm <server> <dependencyId> | Remove a composed dependency (id from deps list). | | mcp servers env | Manage server runtime environment bindings (encrypted secrets injected by the proxy). | | mcp servers env get <server> <bindingName> | Show metadata for one env binding. Plaintext is never returned — values are write-only. | | mcp servers env list <server> | List a server's env bindings (metadata only; values are never returned). | | mcp servers env set <server> <bindingName> [value] | Create or replace an env binding's value. Schedules a Cloudflare-runtime push when the server has an active deployment. | | mcp servers env unset <server> <bindingName> | Remove an env binding. Idempotent — succeeds even if the binding is already gone. | | mcp servers export <server> | Export a self-host bundle (Dockerfile + node entry + config) for a server | | mcp servers generate <server> | Generate the server's TypeScript bundle (the exact code deploy ships) and optionally write it to disk — a dry-run for the codegen path. Wraps POST /api/v1/server/generate. | | mcp servers get <server> | Get details for a single server (accepts id, name, or slug) | | mcp servers ingest | Create a server from an OpenAPI spec (file, URL, or @- stdin) or a GraphQL endpoint. Wraps POST /api/v1/server/ingest. | | mcp servers list | List servers | | mcp servers logs <server> | Show recent deployment events for a server's latest deployment | | mcp servers oauth-connections <server> | List OAuth provider connections bound to this server | | mcp servers pause <server> | Pause the server's latest active deployment (traffic stops; worker is preserved) | | mcp servers prompts | Author the prompts a client reads as this server’s workflows (list / set / rm) | | mcp servers prompts list <server> | List the prompts authored on a server. | | mcp servers prompts rm <server> <name> | Remove a prompt by name. | | mcp servers prompts set <server> <name> | Create or replace a prompt by name. Re-running with the same name updates it. | | mcp servers push-spec <server> | Regenerate a server’s bundle from a local OpenAPI spec. Previews the tool changes; --apply writes them. | | mcp servers regenerate <server> | Re-fetch a server's source spec and rebuild its tools with the current codegen (preserves customizations). Wraps POST /api/v1/server/regenerate. | | mcp servers resources | Author the resources this server exposes to clients (list / set / rm) | | mcp servers resources list <server> | List the resources authored on a server. | | mcp servers resources rm <server> <uri> | Remove a resource by URI. | | mcp servers resources set <server> <uri> | Create or replace a resource by URI. Re-running with the same URI updates it. | | mcp servers resume <server> | Resume the server's paused deployment (flips dispatch routing back on; only platformWorker deployments support instant resume) | | mcp servers score <server> | Score whether an agent handed this catalog picks the right tool (starts a run, or reads the last one) | | mcp servers shape <server> | Propose task-shaped tools over this server’s endpoints (starts a run, or reads the last one) | | mcp servers test-runs | Inspect and trigger server sandbox test runs | | mcp servers test-runs get <testRunId> | Show one server test run with assertions, trace, and the tool response | | mcp servers test-runs list <server> | List sandbox test runs for a server | | mcp servers test-runs new <server> | Trigger a sandbox test run against a server's latest deployment (synchronous) | | mcp servers test-scenarios | Inspect server sandbox test scenarios | | mcp servers test-scenarios generate <server> | Seed one baseline smoke-test scenario per tool from the tools’ schemas (safe to re-run) | | mcp servers test-scenarios get <scenarioId> | Show one server test scenario including its request and expected response | | mcp servers test-scenarios list <server> | List test scenarios bound to a server | | mcp servers test-scenarios new <server> | Author a test scenario for one of the server’s tools | | mcp servers test-scenarios rm <server> <scenarioId> | Archive a test scenario | | mcp servers test-scenarios update <server> <scenarioId> | Update a test scenario (only the flags you pass change) | | mcp servers test-suites | Inspect server sandbox test suites (groups of scenarios) | | mcp servers test-suites get <suiteId> | Show one server test suite including its scenario membership | | mcp servers test-suites list <server> | List test suites bound to a server | | mcp servers test-suites run <server> | Run the server’s test scenarios and report the suite result | | mcp servers update <server> | Update a server (name, description, tool surface, upstream key mode, runtime token TTL) | | mcp servers variables | Inspect and set the sandbox variables that scenario arguments reference | | mcp servers variables list <server> | List a server’s sandbox variables and which ones still need a value | | mcp servers variables set <server> | Set or clear sandbox variables (repeat --set; an empty value clears) | | mcp servers version <server> <versionId> | Show details for a single server version | | mcp servers versions <server> | List published versions of a server |

mcp skills — Manage skills

| Command | Description | | --- | --- | | mcp skills archive <skill> | Archive a skill (replay-as-success on already-archived rows) | | mcp skills connect <skill> | Print (or apply) the MCP configuration that connects this skill to your coding agent | | mcp skills create | Create an empty skill draft in a project | | mcp skills delete <skill> | Permanently delete a draft skill. Requires --confirm "". Published skills with versions/installs are rejected (archive instead). | | mcp skills deploy <skill> | Deploy a published skill version as an MCP endpoint (newest version by default) | | mcp skills get <skill> | Get details for a single skill (accepts id or name) | | mcp skills install <skill> | Install a skill into the caller's project | | mcp skills installations <skill> | List the installation footprint of a skill (which orgs / users installed it) | | mcp skills invoke <skill> | Run a deployed skill from the terminal. Resolves the skill’s MCP endpoint, auto-discovers its single tool, and calls it. | | mcp skills list | List skills in an organization | | mcp skills publish <skill> | Cut an immutable version of a skill | | mcp skills pull <registryArtifactId> | Download an installed skill as SKILL.md + mcpcloud.json for your coding agent | | mcp skills set-content <skill> | Write a skill’s authored content: instructions, dependencies, guardrails, schemas, composition graph | | mcp skills test-runs | Inspect and trigger skill sandbox test runs | | mcp skills test-runs get <testRunId> | Show one skill test run with assertions, trace, and output | | mcp skills test-runs list <skill> | List sandbox test runs for a skill | | mcp skills test-runs new <skill> | Trigger a sandbox test run against a skill (synchronous; returns terminal status) | | mcp skills test-scenarios | Inspect skill sandbox test scenarios | | mcp skills test-scenarios generate <skill> | Seed suggested scenarios from the skill’s examples and its dependencies’ tools (safe to re-run) | | mcp skills test-scenarios get <scenarioId> | Show one skill test scenario including its prompt and assertion body | | mcp skills test-scenarios list <skill> | List test scenarios bound to a skill | | mcp skills test-scenarios new <skill> | Author a sandbox test scenario for a skill | | mcp skills test-scenarios rm <skill> <scenarioId> | Archive a skill test scenario | | mcp skills test-scenarios update <skill> <scenarioId> | Update a skill test scenario (only the flags you pass change) | | mcp skills token <skill> | Reveal the skill’s runtime bearer token (--rotate issues a new one and invalidates the old) | | mcp skills uninstall <skill> | Mark the caller's installation of this skill as removed | | mcp skills update <skill> | Update skill metadata (name, description, visibility) | | mcp skills version <skill> <versionId> | Show details for one published skill version (workflow summary, triggers, deps) | | mcp skills versions <skill> | List published versions of a skill |

mcp tools — Inspect and edit tools on a project or server

| Command | Description | | --- | --- | | mcp tools diff <name> | Compare a tool's local files (metadata .md + handler .ts) against the cloud version. | | mcp tools edit <name> | Open a tool's local .md in $EDITOR; pushes your changes to the cloud on save. Requires mcp dev to have materialized tools (run mcp tools pull first if needed). | | mcp tools enrich [name] | Run AI enrichment for one tool (prints the suggested vs applied diff), or for the whole server with --all. | | mcp tools handlers | Inspect tool handler files (the per-tool .ts in src/tools/) | | mcp tools handlers list | List every tool with its handler override status (default / override / stale). | | mcp tools list | List tools for a project, or scope to a single server with --server | | mcp tools pull [name] | Refresh local tool files from the cloud, overwriting any unsaved local edits. Use after 409 conflict to re-sync. | | mcp tools pull-handler <name> | Refresh a tool's handler source from the cloud, overwriting any local edits. Use after 409 conflict to re-sync. | | mcp tools reset-handler <name> | Drop the cloud handler override for a tool and re-materialize the codegen default. Prompts to confirm unless --yes is passed. | | mcp tools show <name> | Show a tool's full applied metadata (and any pending AI suggestion). |

mcp usage — Show organization usage (tool-call volume) and credit balance from the terminal

| Command | Description | | --- | --- | | mcp usage ledger | List organization credit ledger entries (most recent first) |

Plugins

mcp plugin install <spec> extends the CLI with third-party commands. A plugin runs in your CLI process with your credentials and the same filesystem access — install only plugins you trust.

Configuration file

~/.mcpcloud/config.json (mode 0600). Environment variables take precedence over its contents. Recognised fields:

{
  "apiKey": "mck_...",                       // saved by `mcp login`
  "baseUrl": "https://your-deployment...",   // saved by `mcp config set-url` (omit to use the production default)
  "defaultOrganizationId": "org_...",        // saved by `mcp config set-org`
  "profiles": { "staging": { /* apiKey + baseUrl + defaultOrganizationId */ } }
}

Reliability

• Timeout: every request is bounded by a 30-second AbortController.
• Retry: transient failures (network errors, HTTP 502/503/504) are retried once with jittered backoff. 4xx responses are never retried.
• Error envelope: server errors carry their code, message, requestId, and docsUrl so you can correlate with backend logs.

Development

git clone https://github.com/MCPCloud-sh/mcp-hub
cd mcp-hub/packages/cli

bun install
bun run dev -- --help            # run from source
bun run typecheck                # tsc --noEmit
bun run test                     # vitest
bun run build                    # bundle to dist/index.js
bun run docs:readme              # regenerate the command reference above

The command reference between the markers is generated from the live Commander tree by bun run docs:readme; a conformance test fails if it drifts, so a new command can't ship with a stale README.

Releasing

The CLI is released independently of the dashboard via Changesets.

Support

• Documentation: https://mcpcloud.sh/docs
• Email: [email protected]

License

MIT © MCPCloud.sh