CLI

JinjaRope contains a simple CLI interface which allows rendering Jinja templates using the JinjaRope Environment.

Very basic usage:

jinjarope -i "input_file.jinja" [-o "output.md"] [-j "additional_jinja_files.toml"] [...]

Since jinjarope is using fsspec in the background, templates can be loaded from all filesystems fsspec understands. This allows to load templates from http:// sources, zip files, specific git revisions and a lot more different sources. The same counts for the output path.

jinjarope -i "github://phil65:mknodes@main/docs/icons.jinja" -o "output.md" -j "path_to_jinja_file.toml"

LLMling Agent

🤖 LLMling Agent CLI - Run and manage LLM agents

(No active config set)

llmling agent [OPTIONS] COMMAND [ARGS]...

--install-completion

Flag Install completion for the current shell.

--show-completion

Flag Show completion for the current shell, to copy it or customize the installation.

--help

Flag Show this message and exit.

add

Add a new agent configuration file.

llmling agent add [OPTIONS] NAME PATH

name

"REQUIRED"

path

"REQUIRED"

--verbose, -v

Flag Enable debug logging

--help

Flag Show this message and exit.

run

Run a node (agent/team) with prompts.

Examples: # Single agent llmling run myagent "Analyze this"

    # Team
    llmling run myteam "Process this"

    # Sequential team
    llmling run mysequence "Process this"

llmling agent run [OPTIONS] NODE_NAME [PROMPTS]...

node_name

"REQUIRED"

prompts

--config, -c

Override config path

--show-messages

Default: True Flag Show all messages (not just final responses)

--detail, -d

Default: simple Output detail level: simple/detailed/markdown

--metadata

Flag Show message metadata

--costs

Flag Show token usage and costs

--verbose, -v

Flag Enable debug logging

--help

Flag Show this message and exit.

list

List agents from the active (or specified) configuration.

llmling agent list [OPTIONS]

--config, -c

Name of agent configuration to list (defaults to active)

--output-format, -o

Default: text Output format. One of: text, json, yaml

--verbose, -v

Flag Enable debug logging

--help

Flag Show this message and exit.

set

Set the active agent configuration file.

llmling agent set [OPTIONS] NAME

name

"REQUIRED"

--verbose, -v

Flag Enable debug logging

--help

Flag Show this message and exit.

chat

Start interactive chat session with an agent.

The agent can forward responses to other agents creating a processing chain. Use --forward-to to specify target agents and --wait-chain to control whether to wait for the full chain to complete.

llmling agent chat [OPTIONS] AGENT_NAME

agent_name

"REQUIRED"

-s, --session-id

Session id

-c, --config

Override agent config path

-m, --model

Override agent's model

--stream

Default: True Flag Enable streaming mode (default: off)

-f, --forward-to

Multiple values allowed. Forward responses to these agents

-l, --log-level

Default: WARNING Log level (DEBUG, INFO, WARNING, ERROR)

--help

Flag Show this message and exit.

quickstart

Start an ephemeral chat session with minimal setup.

llmling agent quickstart [OPTIONS] [MODEL]

model

Default: openai:gpt-4o-mini

-l, --log-level

Default: WARNING Log level (DEBUG, INFO, WARNING, ERROR)

--stream

Default: True Flag Enable streaming mode (default: off)

--help

Flag Show this message and exit.

watch

Run agents in event-watching mode.

llmling agent watch [OPTIONS] CONFIG

config

"REQUIRED"

--show-messages

Default: True Flag Show all messages (not just final responses)

--detail, -d

Default: simple Output detail level: simple/detailed/markdown

--metadata

Flag Show message metadata

--costs

Flag Show token usage and costs

--log-level

Default: INFO Logging level

--help

Flag Show this message and exit.

serve

Run agents as an MCP server.

This makes agents available as tools to other applications, regardless of whether pool_server is configured in the manifest.

llmling agent serve [OPTIONS] CONFIG

config

"REQUIRED"

--transport

Default: stdio Transport type (stdio/sse)

--host

Default: localhost Host to bind server to (sse only)

--port

Default: 8000 Port to listen on (sse only)

--zed-mode

Flag Enable Zed editor compatibility

--show-messages

Flag Show message activity

--log-level

Default: INFO Logging level

--help

Flag Show this message and exit.

task

Execute a task with the specified agent.

llmling agent task [OPTIONS] AGENT_NAME TASK_NAME

agent_name

"REQUIRED"

task_name

"REQUIRED"

-c, --config

Agent configuration file

-p, --prompt

Additional prompt

-l, --log-level

Default: WARNING Log level (DEBUG, INFO, WARNING, ERROR)

--help

Flag Show this message and exit.

ui

Start the UI configured in the manifest.

llmling agent ui [OPTIONS]

--config, -c

Override config path

--help

Flag Show this message and exit.

launch

Launch the web interface.

llmling agent launch [OPTIONS]

-h, --host

Default: 127.0.0.1 Host to bind to

-p, --port

Port to bind to

--share

Flag Create public URL

--theme

Default: soft UI theme (soft/base/monochrome/glass/default)

--browser

Default: True Flag Open in browser

--help

Flag Show this message and exit.

history

Conversation history management

llmling agent history [OPTIONS] COMMAND [ARGS]...

--help

Flag Show this message and exit.