#Public, open-source repository

smbcloud-cli is published on GitHub as open source. Everything you commit here — code, comments, docs, and .agents/skills/* — is world-readable and permanent (git history outlives any later deletion). Before writing anything, ask: "is this safe on a public repo forever?"

Do not add internal smbCloud infrastructure detail or secrets:

• server hostnames/IPs and operational endpoints beyond what the CLI already targets in source — e.g. account-scoped SSH key names (id_<n>@smbcloud), api-1.smbcloud.xyz, internal health ports
• account/user IDs and which user or tenant owns which project
• real customer/app domains, PM2 process names, the production port → app → domain registry
• workspace/project IDs and frontend_app_id / deploy_repo_id values
• secrets/config: API keys, tokens, .env values, connection strings, real auth/CORS origins
• commands that read local credentials to enumerate the API (e.g. cat ~/.smb/token | curl …)
• incident logs or examples that name real apps, tenants, customers, or dated rollouts

Keep docs and skills generic — describe how the tool behaves, using placeholders (example.com, <app>, <source>, <port>, <n>). The base API host (api.smbcloud.xyz) is already in the source, so it is not a new leak; the items above are. Fleet-specific operational detail and the internal deploy reference live only in the private smbcloud repo.

#Rust coding guidelines

• Prioritize code correctness and clarity. Speed and efficiency are secondary priorities unless otherwise specified.
• Do not write organizational or comments that summarize the code. Comments should only be written in order to explain "why" the code is written in some way in the case there is a reason that is tricky / non-obvious.
• Prefer implementing functionality in existing files unless it is a new logical component. Avoid creating many small files.
• Avoid using functions that panic like unwrap(), instead use mechanisms like ? to propagate errors.
• Be careful with operations like indexing which may panic if the indexes are out of bounds.
• Never silently discard errors with let _ = on fallible operations. Always handle errors appropriately:
  • Propagate errors with ? when the calling function should handle them
  • Use .log_err() or similar when you need to ignore errors but want visibility
  • Use explicit error handling with match or if let Err(...) when you need custom logic
  • Example: avoid let _ = client.request(...).await?; - use client.request(...).await?; instead
• When implementing async operations that may fail, ensure errors propagate to the UI layer so users get meaningful feedback.
• Never create files with mod.rs paths - prefer src/some_module.rs instead of src/some_module/mod.rs.
• When creating new crates, prefer specifying the library root path in Cargo.toml using [lib] path = "...rs" instead of the default lib.rs, to maintain consistent and descriptive naming (e.g., gpui.rs or main.rs).
• Avoid creative additions unless explicitly requested
• Use full words for variable names (no abbreviations like "q" for "queue")
• Use variable shadowing to scope clones in async contexts for clarity, minimizing the lifetime of borrowed references. Example: rust executor.spawn({ let task_ran = task_ran.clone(); async move { *task_ran.borrow_mut() = true; } });

#Build, test, and run

This is a Cargo workspace pinned to stable Rust (rust-toolchain.toml). The CLI binary is named smb (crate smbcloud-cli, in crates/cli).

cargo build                              # build everything (debug)
cargo run -p smbcloud-cli -- <args>      # run the CLI, e.g. -- me, -- deploy, -- --help
cargo run -p smbcloud-cli -- -e dev me   # talk to a local API (Environment::Dev -> http://localhost:8088)

CI gate (mirror it locally before pushing — see .github/workflows/ci.yml):

cargo fmt --all -- --check
cargo clippy --workspace --exclude smbcloud-auth-sdk-wasm --tests -- -D warnings   # warnings are denied
cargo test  --workspace --exclude smbcloud-auth-sdk-wasm
cargo test -p smbcloud-model app_auth::tests::some_test    # a single test

Always run cargo fmt --all and the cargo clippy command above before considering any Rust change done — after every edit, not just before a push. Fix clippy findings rather than silencing them with #[allow(...)] unless there's a specific reason the lint doesn't apply, and note that reason inline.

ci.yml runs six jobs on every push; run the ones relevant to what changed, not just the main Rust workspace job — a change that only compiles in the main workspace but breaks one of these is still a broken CI run:

The last four jobs need their own toolchain (maturin/PyO3, wasm-pack + Node, the .NET SDK, Ruby/bundler) — if it isn't installed, say so explicitly rather than skipping the check silently, so the user knows it wasn't verified.

smbcloud-auth-sdk-wasm targets wasm32-unknown-unknown and cannot build or test on the host — it is always excluded from workspace clippy/test and checked separately: cargo check -p smbcloud-auth-sdk-wasm --target wasm32-unknown-unknown.

CLI_CLIENT_SECRET (see .env.example) is the OAuth client-credentials secret read at runtime; tests and cargo check don't need it, but login flows against a real API do.

#Releases

Releases are prepared on the development branch with a clean tree via the Makefile, which drives cargo workspaces to bump every crate in lockstep, then syncs the package-manager manifests (npm/nuget/pypi) and regenerates the SDK lockfiles:

make patch | make minor | make major | make custom VERSION=0.x.y

It commits Release <version> and tags v<version> locally; pushing is manual. Crate publishing is separate and manual (cargo workspaces publish, see docs/development.md). The per-target release workflows (release-*.yml) build the distributable artifacts.

server.json and server-xcrs.json (the MCP Registry listings) are version-synced by the same Makefile step, and release-nuget.yml triggers release-mcp-registry.yml once its publish job succeeds — the listings need the npm, NuGet, and Cargo packages live at their recorded versions, and the registry validates their metadata against the real packages. See docs/mcp-registry.md.

#Architecture

smb is a thin async (tokio) clap front-end over a set of smbcloud-* library crates that talk to the smbCloud REST API. The product purpose is deploying apps (Node.js/Next.js, Ruby/Rails, Rust, Swift/Vapor, static/Vite) from the terminal.

#Request flow

crates/cli/src/main.rs → cli/mod.rs (the Cli/Commands clap tree) → a process_* handler under account/, project/, deploy/, or mail/. Each handler returns a CommandResult (a spinner + final symbol/message); main prints it and sets the exit code. The top-level --environment (dev/production, default production) threads through every handler and selects the API host and the on-disk state dir (.smb vs .smb-dev, see smbcloud-network/src/environment.rs).

Two cross-cutting globals are resolved once in main before any handler runs: - CI mode (crates/cli/src/ci.rs): --ci / SMB_CI=1 / conventional CI env. Stored in a process-global AtomicBool (ci::is_ci()) rather than threaded through signatures, so deep prompt code can fail fast instead of blocking on a missing TTY. New interactive prompts must check is_ci() and use ci::interactive_message(...). - Logging: bunyan JSON to ~/<smb_dir>/smbcloud-cli.log (not stdout); user-facing output is console/spinner styling.

#Crate layout (workspace)

• smbcloud-model — shared serde types (Project, User, Runner, deploy config, ErrorCode/ErrorResponse). crate-type = ["cdylib","rlib"] because the SDK crates re-export it. Most unit tests live here.
• smbcloud-network — Environment enum (host/protocol/state-dir resolution) and connectivity checks.
• smbcloud-networking — SmbClient (the OAuth client identity) and the low-level HTTP client/constants.
• smbcloud-networking-project — typed REST calls for projects, deploy repos, frontend apps, deploy config, and deployments (the crud_* files).
• smbcloud-auth / smbcloud-auth-sdk — auth flows (login, signup, OIDC, Apple, client-credentials, reset/verify). auth is the in-CLI implementation; auth-sdk is the embeddable client.
• smbcloud-auth-sdk-wasm / -py (+ sdk/) — language bindings: WASM/wasm-bindgen for npm, PyO3 cdylib for Python, and a separate rb_sys Cargo workspace under sdk/gems for the Ruby gem. These wrap smbcloud-auth-sdk.
• smbcloud-mail, smbcloud-gresiq(-sdk), smbcloud-s6n — clients for the Mail, GresIQ (serverless Postgres), and S6n (S3-compatible storage) products.
• smbcloud-utils — .smb/config.toml (Config) parsing and SSH key-path resolution.

#Deploy subsystem (crates/cli/src/deploy/)

The most involved area. process_deploy.rs loads .smb/config.toml (running interactive setup_project if missing), overlays server-side config, validates project access, then dispatches by config.project.kind to a per-stack path: vite-spa and sveltekit (SvelteKit with adapter-static) share process_deploy_static_site.rs; nextjs-ssr, rails, rust, swift each have their own process_deploy_*.rs. If kind is unset it falls back to deployment_method: Rsync (build locally, upload over rsync, restart over SSH) or Git (push to a remote git hook), with detect_runner.rs sniffing the framework from package.json/Gemfile/Package.swift/Cargo.toml.

Monorepo: runner = Monorepo with a [[projects]] array in config; smb deploy --project <name> (or an interactive picker) swaps config.project to the named sub-project before the dispatch above. migrate.rs/smb migrate pushes local deploy fields up to the server.

When editing a deploy path, keep the public-repo policy above in mind: SSH key names, hosts, ports, app/tenant identifiers, and PM2 process names must stay as placeholders — concrete fleet detail belongs only in the private smbcloud repo.