@amaster.ai/runtime-cli 1.1.13 → 1.1.14-beta.1

package/dist/skill/SKILL.md

@@ -149,6 +149,35 @@ amaster entity bulk-update default users --app myapp --items '[{"id":"123","stat
 amaster entity bulk-delete default users --app myapp --ids '123,456'
 ```
 
+### Model Discovery
+
+```bash
+amaster model datasources [--app <app-code>]
+amaster model table-schema <source-id> [--app <app-code>] [--tables '<csv-or-json-array-or-@file>']
+```
+
+Command semantics:
+
+- `datasources`: discover runtime data sources and the models/tables under each source.
+- `table-schema <source-id>`: fetch `CREATE TABLE` statements for one datasource, optionally narrowed to selected tables.
+
+Important details:
+
+- These commands use the runtime-facing `/api/entity/source/...` endpoints, not the internal MCP `innerapi` routes.
+- Use `datasources` first to learn the real `sourceId`; do not guess it.
+- `--tables` accepts comma-separated names, a JSON array, or `@file`.
+- `table-schema` returns all table schemas in the datasource when `--tables` is omitted.
+- This is the current supported way for an agent using `amaster` to discover available tables before running `entity` queries.
+- Real validated example: `https://bleulig7o.helige.cn/api/entity/source/list_with_models` returns datasource `source-ae0rbo4tg9a8`, and `table-schema` succeeds for tables such as `products`, `categories`, and `stock_out_applications`.
+
+Minimal examples:
+
+```bash
+amaster model datasources --app bleulig7o
+amaster model table-schema source-ae0rbo4tg9a8 --app bleulig7o
+amaster model table-schema source-ae0rbo4tg9a8 --app bleulig7o --tables products,categories
+```
+
 ### BPM
 
 ```bash
@@ -258,17 +287,24 @@ amaster bpm user-roles user_2 --app myapp
 ### Workflow
 
 ```bash
+amaster workflow list [--app <app-code>] [--page <n>] [--limit <n>] [--mode <mode>]
+amaster workflow get <id> [--app <app-code>]
 amaster workflow run <name> [--app <app-code>] [--input '<json-or-@file>'] [--request '<json-or-@file>'] [--response-mode <mode>] [--user <user>] [--files '<json-or-@file>'] [--trace-id <traceId>]
 ```
 
 Command semantics:
 
+- `list`: discover workflow apps exposed by the current runtime app.
+- `get <id>`: fetch one workflow DSL and derive its start-node input variables.
 - `run <name>`: call `workflow-client.run` and print the workflow run result.
 
 Important details:
 
-- `name` must be the workflow YAML filename without extension and without subpaths.
-- Workflow names are app-specific. The current SDK and CLI do not provide a `workflow list` command, so agents must obtain the workflow name from app docs, UI, or the operator.
+- `list` defaults to `mode=workflow`, because `workflow run` only applies to runnable workflow apps.
+- `list` returns two different identifiers: use `id` with `amaster workflow get <id>`, and use `runnable_name` with `amaster workflow run <runnable_name>`.
+- `get <id>` expects the workflow app ID from `workflow list`, not the runnable name.
+- `get <id>` returns `name`, and that `name` is the runnable workflow name for `workflow run`.
+- `name` in `workflow run` must be the runnable workflow name without subpaths.
 - `--input` must be a JSON object and maps to `WorkflowRunRequest.inputs`.
 - `--request` accepts the full `WorkflowRunRequest` object.
 - Explicit flags such as `--input`, `--response-mode`, `--user`, `--files`, and `--trace-id` override the same keys inside `--request`.
@@ -281,6 +317,8 @@ Important details:
 Minimal examples:
 
 ```bash
+amaster workflow list --app myapp --format pretty
+amaster workflow get bdcb98cf-613b-4ad7-bd41-d9a7f8aec199 --app rcz7v42lr --format pretty
 amaster workflow run data-processor --app myapp --input '{"documentId":"abc"}'
 amaster workflow run data-processor --app myapp --request '{"inputs":{"documentId":"abc"},"user":"user-1"}'
 amaster workflow run text-sentiment-analysis --app rcz7v42lr --input '{"input_text":"I love this product. It is fast and easy to use."}'
@@ -318,7 +356,6 @@ amaster s3 metadata uploads/report.pdf --app myapp
 ## Scope Limits
 
 - Only use commands documented in this skill. Do not invent additional subcommands from intuition, stale README text, or helper functions that are not wired into the public CLI.
-- Workflow discovery is not available in the current CLI or SDK surface. There is no `amaster workflow list`, so workflow names must come from app docs, UI, or the operator.
 - Local app config is stored in `~/.amaster/config.json`.
 
 ## Runtime State And File Locations

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@amaster.ai/runtime-cli",
-  "version": "1.1.13",
+  "version": "1.1.14-beta.1",
   "description": "CLI for Amaster SDK - Multi-app support for OpenClaw",
   "type": "module",
   "main": "./dist/index.cjs",
@@ -39,7 +39,7 @@
     "ora": "^8.0.1",
     "yaml": "^2.3.4",
     "inquirer": "^9.2.12",
-    "@amaster.ai/client": "1.1.13"
+    "@amaster.ai/client": "1.1.14-beta.1"
   },
   "devDependencies": {
     "@types/inquirer": "^9.0.7",