cloudbase-declarative-deploy skill

CloudBase Declarative Deploy

Deploy a whole CloudBase project from one cloudbaserc config as desired state, using the deployBuild (local hosting build), deployPlan (dry-run) and deployApply (apply) MCP tools. The orchestrator applies resources in a fixed dependency order:

database → functions → app → hosting → gateway

Sibling skills (local only)

Sibling CloudBase skills ship beside this skill. Use local relative paths such as ../cloudbase-cli/SKILL.md.

Cloud-hosted MCP mode does not guarantee access to a local workspace filesystem or stable relative paths. If a referenced sibling file is not available in cloud mode, use this skill's embedded guidance as source of truth and ask the user for any missing constraints (or to install the missing skill). Do not HTTP-fetch remote skill or protocol markdown into the agent context.

Cross-cutting protocols (required before applying any deploy):

• Change Safety Protocol: ../cloudbase-platform/references/protocols/change-safety-protocol.md
• Deployment Gate: ../cloudbase-platform/references/protocols/deployment-gate.md

When to use this skill

• The project has a cloudbaserc.json / .yaml / .yml / .js describing multiple resources, and the user wants to deploy them together as one config.
• The user asks for 声明式部署 / 配置式部署 / "deploy from cloudbaserc" / "deploy the whole project".
• The user wants to preview what a deploy will change before applying (dry-run plan).
• The user wants to build the static hosting artifact locally first (declarative hosting deploys no longer build implicitly — see deployBuild).
• Multi-environment deploy: production/staging via mode + envOverrides.

Do NOT use for

• Deploying a single cloud function or one static site via tcb CLI → ../cloudbase-cli/SKILL.md.
• In-app SDK integration (web/miniprogram/node) → the matching SDK skill.
• Console UI operations.

Cloud mode: deploy through the two-step upload channel

deployBuild / deployPlan / deployApply in this skill are the local-form declarative executor. In cloud-hosted MCP mode they are intentionally not registered, because that runtime has no local cwd and no filesystem-bound execution path.

This is an execution-channel switch, not a capability gap. Everything still deploys, but the artifact is uploaded first and the build runs in the cloud. If those three tools are missing from the tool list, that is expected — pick the channel that matches the resource.

Cloud functions — two-phase zip

1. queryFunctions(action=getFunctionUploadUrl, functionName?) returns uploadUrl, uploadHeaders, cosBucketName, cosBucketRegion, cosObjectName, expiresInSeconds
2. PUT the function zip to uploadUrl with uploadHeaders
3. manageFunctions(action=createFunction|updateFunctionCode, code={cosBucketName, cosObjectName, cosBucketRegion?}) — phase B deploys from the environment's own COS bucket (deployMode=cos); pass the triple through unchanged, the schema is strict

contentPath (a local directory) is rejected in cloud mode. For layer contents use base64Content; for code use the two-phase path above.

CloudApp — two-step upload plus a cloud pipeline

1. queryApps(action=getUploadUrl, serviceName=...) returns uploadUrl, uploadHeaders, unixTimestamp
2. Upload the source/build zip to uploadUrl with the returned headers (Content-Type: application/zip)
   • Public dependencies resolvable from a lockfile: package.json plus lockfile is enough
   • Private or offline dependencies: package node_modules explicitly
3. manageApps(action=deployApp, cosTimestamp=<unixTimestamp>, installCmd?, buildCmd?, deployCmd?)
   • installCmd / buildCmd / deployCmd are pipeline declarations executed in the cloud container; you pass data and declarations, you do not run local shell commands

Static hosting — route through CloudApp

manageHosting(action=upload) is blocked in cloud mode because it needs a local path. Deploy a static site through the CloudApp two-step upload above (upload the prebuilt artifact, exclude node_modules). Uploading or downloading an already-hosted file has no remote substitute and needs local mode. The other hosting actions (delete, setWebsiteDocument, enableService, bindDomain, unbindDomain, updateDomain) remain available.

What cloud mode does not give you

• No dry-run. There is no deployPlan against a local cloudbaserc. Do not claim a plan was generated or reviewed in cloud mode; state what is about to be deployed and what the upload contains. Roadmap: upload cloudbaserc as a data artifact, then run server-side plan/apply orchestration.
• No local project context. mode / envOverrides resolution from a project file does not apply; pass the target environment explicitly.

For parameter details, see references/plan-and-apply.md (Cloud-hosted upload pipeline path).

Core principles

1. Plan before apply — always. Run deployPlan first (dry-run, zero side effects). Read the per-resource action classification and show it to the user before calling deployApply.

2. Apply requires explicit confirm. deployApply will refuse unless confirm=true is passed. This is the destructive-write guard.

3. Deployment Gate. Before any apply, complete cloudbase-platform/references/protocols/deployment-gate.md and present the mandatory declaration.

4. Conservative on existing resources by default. yes defaults to false → existing resources are skipped, not overwritten. Only pass yes=true when the user explicitly wants to overwrite/update existing resources.

5. database failure always aborts. Even with continueOnError=true, a database-stage failure stops the whole deploy, because later resources depend on it.

6. Resolve envId explicitly. Never rely on implicit defaults silently — know which environment is targeted (see the priority table below) and confirm it with the user before applying.

Plan action classification

deployPlan returns a list of resource entries. Each status means:

status	meaning
create	new resource, will be created
update	exists, will be overwritten/updated
skip	no change needed
conflict	conflict detected — deploy will abort, must resolve first
deploy	direct overwrite upload

If any entry is conflict, stop and resolve it before applying.

envId resolution priority

explicit envId param  >  cloudbaserc `envId`  >  logged-in / bound environment

If none can be resolved, the tool errors out. Prefer confirming the resolved envId with the user before applying to production.

Local workflow (build → plan → apply)

When app or hosting needs built artifacts, declarative deploy is a three-step flow — deployApply runs no build for either resource type:

1. Ensure a cloudbaserc config exists under the project root (cwd).
2. Build first. Which executor depends on the resource:
   • hosting with a buildCommand: call deployBuild({ cwd, mode? }) to produce the local artifacts (builds every hosting item; pure-static items without a build command are skipped automatically).
   • app: build it yourself in the project directory — deployBuild does not cover app. Run deployPlan first and use buildHint.buildCommand / buildHint.outputDir to know what to run and where the artifacts must land.
   • If the build artifacts are missing at apply time, deployApply fails with BUILD_OUTPUT_NOT_FOUND and directs you back to this step.
   • deployBuild needs no envId and no confirm (local build only, never touches cloud resources); if dependencies are not installed it fails with DEPENDENCY_NOT_INSTALLED and tells you to run install first.
3. Call deployPlan (optionally with mode, envId, only, skip). Read the plan, and check buildHint.artifactReady on every app / hosting item.
4. Present the plan + Deployment Gate declaration to the user; get confirmation.
5. Call deployApply with confirm=true (plus yes / concurrency / continueOnError as needed). Reuse the same mode / envId / only / skip as the plan. A hosting item with existing build output is uploaded directly (the tool clears the build command and reports hostingNeutralized: true); rebuild after source changes so the upload is not stale.
6. Report the applied result back to the user.

For cloud-hosted MCP mode, do not ask for local cwd/filesystem reads; use the Cloud mode upload-channel flow above.

Build & pipeline execution

Use this mental model: build → plan → apply. The build executor depends on resource type.

deployApply never runs a build — neither for app nor for hosting. buildCommand / installCommand / framework in cloudbaserc are only used to infer the output directory; they are never executed by apply.

resource	typical build executor	deployment path
hosting	deployBuild (local shell build, run before apply)	deployApply uploads the built output directly; missing output → BUILD_OUTPUT_NOT_FOUND
app	you build it — run the build in the project directory before apply (deployBuild does not cover app)	deployApply uploads outputDir as-is; missing/empty → BUILD_OUTPUT_NOT_FOUND
app via cloud pipeline	CLI tcb app deploy (source zip + cloud build)	not reachable through deployApply — use the CLI when a cloud build is wanted
functions	local zip / cloud build / image pipeline	depends on buildStrategy (zip/cloud/local/image)

Reading the build context from deployPlan

app and hosting plan items carry a machine-readable buildHint:

"buildHint": {
  "buildCommand": "npm run build",  // null → nothing to build
  "outputDir": "/abs/path/to/dist", // what apply will upload
  "artifactReady": false            // false → apply will fail
}

Read these fields instead of parsing the localized text in details. When artifactReady is false: run deployBuild for hosting, or execute buildHint.buildCommand in the project directory for app; then re-run deployPlan and confirm artifactReady: true. Never edit the config to work around it.

deployBuild builds every hosting item that has a buildCommand (framework mapping or package.json auto-detection); it skips pure-static items and does not touch app. Build failures surface as BUILD_FAILED, missing local dependencies as DEPENDENCY_NOT_INSTALLED (run install first — deployBuild never installs dependencies for you).

Build command resolution follows declaration priority:

explicit config > framework mapping defaults > package.json auto-detection

buildCommand / installCommand / deployCmd are declarative intent in config. Execution ownership depends on path:

• local-form paths: specific steps may run in local shell executor
• cloud-hosted paths: commands are executed by cloud pipeline container (staticCmd), or replaced by prebuilt artifact upload

So the answer to "can cloud mode run local CLI commands" is: execution authority is moved from local shell to cloud pipeline; agent transmits declarations and artifacts.

Routing

User task	Read
cloudbaserc resource fields & desired-state config shape	references/config-schema.md
deployPlan → deployApply two-step flow, parameters, safety	references/plan-and-apply.md
Multi-env (mode / envOverrides), envId priority, env vars	references/multi-env.md

Minimum self-check

• Built the hosting artifacts first (deployBuild) when hosting declares a buildCommand?
• Built the app artifacts yourself (deployBuild does not cover app) and confirmed buildHint.artifactReady: true in the plan?
• Ran deployPlan and read the action classification before deployApply?
• Resolved and confirmed the target envId?
• Completed the Deployment Gate declaration before applying?
• Passed confirm=true only after user confirmation?
• Left yes=false unless overwrite of existing resources was explicitly requested?
• Handled any conflict entries before applying?

Reference index

All packaged reference files (required for skill lint reachability):

• config-schema.md
• plan-and-apply.md
• multi-env.md