@runtypelabs/cli 2.25.3 → 2.28.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +101 -2
- package/dist/index.js +5722 -2778
- package/package.json +10 -7
package/README.md
CHANGED
|
@@ -27,6 +27,26 @@ Or run without installing:
|
|
|
27
27
|
npx @runtypelabs/cli@latest <command>
|
|
28
28
|
```
|
|
29
29
|
|
|
30
|
+
### Automatic updates
|
|
31
|
+
|
|
32
|
+
Global installations check npm for a new stable release once per hour. When one is available,
|
|
33
|
+
the CLI installs the exact version in the user-owned `~/.runtype/cli/versions/` cache, verifies
|
|
34
|
+
that it starts and reports the expected version, and activates it atomically for the next
|
|
35
|
+
invocation. It never rewrites your npm, pnpm, or yarn global prefix while a command is running.
|
|
36
|
+
|
|
37
|
+
```bash
|
|
38
|
+
runtype update # Check and install now
|
|
39
|
+
runtype update --check # Check without installing
|
|
40
|
+
runtype update --rollback # Return to the previous healthy version
|
|
41
|
+
```
|
|
42
|
+
|
|
43
|
+
Automatic installation is disabled for CI, source checkouts, project-local installations, and
|
|
44
|
+
`npx`/`pnpm dlx` runs so pinned and ephemeral executions remain reproducible. Set
|
|
45
|
+
`RUNTYPE_DISABLE_AUTO_UPDATE=true` to keep using the package-manager-installed version (including
|
|
46
|
+
an intentionally pinned global version), or
|
|
47
|
+
`RUNTYPE_DISABLE_UPDATE_CHECK=true` to keep the active managed version without checking for more
|
|
48
|
+
updates.
|
|
49
|
+
|
|
30
50
|
## Start Here
|
|
31
51
|
|
|
32
52
|
The easiest way to try Runtype starting at the CLI is to run a Marathon task. If you are not logged in yet, the CLI will guide you through login or signup first.
|
|
@@ -66,7 +86,54 @@ runtype login
|
|
|
66
86
|
runtype login --api-key <key>
|
|
67
87
|
```
|
|
68
88
|
|
|
69
|
-
### 2.
|
|
89
|
+
### 2. Connect Your Coding Agent to Runtype MCP
|
|
90
|
+
|
|
91
|
+
Use the CLI in this first session, then install the durable hosted MCP connection for your current
|
|
92
|
+
coding agent:
|
|
93
|
+
|
|
94
|
+
```bash
|
|
95
|
+
runtype install-mcp
|
|
96
|
+
```
|
|
97
|
+
|
|
98
|
+
The command detects Claude Code, Codex, Cursor, Gemini CLI, OpenCode, or VS Code; installs the
|
|
99
|
+
matching Runtype agent skill; configures the user-level hosted MCP server; and starts client-owned
|
|
100
|
+
browser OAuth when the client supports it. Follow its final instruction to restart or reload the
|
|
101
|
+
coding agent. After restart, prefer Runtype MCP and use the CLI as a fallback rather than as the
|
|
102
|
+
primary integration.
|
|
103
|
+
|
|
104
|
+
If several coding agents are installed and the current one cannot be identified from its
|
|
105
|
+
environment, select one explicitly (for example, `runtype install-mcp --agent codex`). Use
|
|
106
|
+
`--dry-run` to inspect the plan and `--force` only to replace a conflicting server already named
|
|
107
|
+
`runtype`.
|
|
108
|
+
|
|
109
|
+
You can also discover and call either hosted MCP server directly from the shell without registering
|
|
110
|
+
individual tool commands:
|
|
111
|
+
|
|
112
|
+
```bash
|
|
113
|
+
# Primary MCP (default)
|
|
114
|
+
runtype mcp tools
|
|
115
|
+
runtype mcp inspect create_flow
|
|
116
|
+
runtype mcp call list_flows -F limit=20
|
|
117
|
+
|
|
118
|
+
# Complete JSON input from a file or stdin
|
|
119
|
+
runtype mcp call create_flow --input arguments.json --json
|
|
120
|
+
cat arguments.json | runtype mcp call create_flow --input - --json
|
|
121
|
+
|
|
122
|
+
# Code Mode MCP
|
|
123
|
+
runtype mcp tools --server code
|
|
124
|
+
runtype mcp call execute --server code -F code=@query.js --json
|
|
125
|
+
```
|
|
126
|
+
|
|
127
|
+
`-f key=value` always sends a string. `-F key=value` parses JSON scalar, array, and object values;
|
|
128
|
+
prefix the value with `@` to read it from a file or use `@-` for stdin. Use `--json-body` for an
|
|
129
|
+
inline complete arguments object. The `call` command sends the provided name directly to MCP, so
|
|
130
|
+
new server tools work without a CLI release. Tool calls wait up to 930 seconds by default so
|
|
131
|
+
long-running flows can complete; use `--timeout <ms>` to choose a different deadline. The separate
|
|
132
|
+
`runtype tools` command manages saved runtime tools. Run `runtype auth login` first or set
|
|
133
|
+
`RUNTYPE_API_KEY` before using `--input -` or `-F key=@-`; the CLI will not put an interactive login
|
|
134
|
+
prompt on a stream that contains tool arguments.
|
|
135
|
+
|
|
136
|
+
### 3. Work With Agents, Flows, and Records
|
|
70
137
|
|
|
71
138
|
```bash
|
|
72
139
|
# List available agents
|
|
@@ -88,7 +155,7 @@ runtype dispatch --flow <flow-id> --variable customerName=Alyss --variable prior
|
|
|
88
155
|
runtype records create --name "My Record" --type "document"
|
|
89
156
|
```
|
|
90
157
|
|
|
91
|
-
###
|
|
158
|
+
### 4. Run Multi-Session Agent Tasks
|
|
92
159
|
|
|
93
160
|
Use `runtype marathon` to run long-running tasks with real-time streaming output. If you want to try it without installing first, use `npx @runtypelabs/cli@latest marathon ...`. Swap the agent, model, tools, and execution environment depending on whether you want to research, edit code, or build and ship something end-to-end.
|
|
94
161
|
|
|
@@ -682,6 +749,38 @@ View analytics and execution results.
|
|
|
682
749
|
| `analytics stats` | Show account statistics |
|
|
683
750
|
| `analytics results` | List execution results (`--flow`, `--record`, `--status`, `--limit`) |
|
|
684
751
|
|
|
752
|
+
### `runtype update`
|
|
753
|
+
|
|
754
|
+
Manage CLI updates.
|
|
755
|
+
|
|
756
|
+
| Command | Description |
|
|
757
|
+
| ------------------- | --------------------------------------------- |
|
|
758
|
+
| `update` | Check npm and install the latest stable CLI |
|
|
759
|
+
| `update --check` | Check for a newer version without installing |
|
|
760
|
+
| `update --rollback` | Activate the previous healthy managed version |
|
|
761
|
+
|
|
762
|
+
### `runtype install-mcp`
|
|
763
|
+
|
|
764
|
+
Install the current coding agent's Runtype skills and user-level hosted OAuth MCP
|
|
765
|
+
connection. Use `--agent <id>` when detection is ambiguous, `--dry-run` to inspect the
|
|
766
|
+
plan, and `--force` only to replace a conflicting user-level server named `runtype`.
|
|
767
|
+
Follow the final OAuth and restart instructions before expecting MCP tools in the running
|
|
768
|
+
agent session.
|
|
769
|
+
|
|
770
|
+
### `runtype mcp`
|
|
771
|
+
|
|
772
|
+
Use the hosted MCP servers directly from scripts and terminals. `mcp tools` lists the live catalog,
|
|
773
|
+
`mcp inspect <name>` shows the current input/output schemas, and `mcp call <name>` invokes any tool
|
|
774
|
+
name without local registration. The standard MCP is the default; pass `--server code` for Code
|
|
775
|
+
Mode. `--json` preserves the complete MCP result for piping into tools such as `jq`.
|
|
776
|
+
|
|
777
|
+
The command follows `--api-url` for the standard endpoint and selects the matching production,
|
|
778
|
+
staging, or local Code Mode host. Local development can override endpoint inference with
|
|
779
|
+
`RUNTYPE_MCP_URL` and `RUNTYPE_CODE_MCP_URL`.
|
|
780
|
+
|
|
781
|
+
Tool calls default to a 930-second client timeout, slightly above Runtype's maximum flow runtime.
|
|
782
|
+
Pass `--timeout <ms>` to use a different positive millisecond deadline.
|
|
783
|
+
|
|
685
784
|
## Configuration
|
|
686
785
|
|
|
687
786
|
```bash
|