How to Install and Use Larry in OpenClaw

Open your terminal and run the clawhub install command to add Larry to your OpenClaw instance. ClawHub downloads the larry skill package from the registry, verifies the download, and registers it with your local OpenClaw configuration. Installation takes 10-20 seconds. When complete, the terminal displays a confirmation with the installed version number and lists any configuration keys that need to be set before the skill fully activates. Some versions of Larry may work with zero additional configuration for basic task automation features, while more advanced features like calendar integration or external task system sync may require additional credentials. Verify the install succeeded by running `clawhub list` — larry should appear in the output. Then run `clawhub info larry` to see the full configuration reference and understand which features require additional setup.

1# Install larry
2clawhub install larry
3
4# Verify installation
5clawhub list
6
7# View full configuration reference
8clawhub info larry

Larry's configuration lives in your OpenClaw `.env` file or in the skill-specific YAML configuration file at `~/.openclaw/skills/larry.yaml`. Unlike some skills that require a third-party API key to function at all, Larry's core features may be usable immediately after installation with sensible defaults. For basic task sequence automation and productivity prompts, review the output of `clawhub info larry` to understand what settings are available. Common configuration options for utility skills like Larry include default task format preferences, session history retention length, and integration hooks for external tools. If Larry requires any configuration variables (such as credentials for an integrated external service), add them using `clawhub config set VARIABLE_NAME value` and then run `clawhub reload` to apply. If Larry works without additional configuration, you can proceed directly to testing — the next step will confirm the skill is active. The larry.yaml skill configuration file is where you define any persistent automation patterns or default workflow templates that you want Larry to recognize without you describing them from scratch each time.

1# Check if any config is needed (review the output)
2clawhub info larry
3
4# Set any required config values (replace with actual variable names from info output)
5clawhub config set LARRY_CONFIG_KEY your_value_here
6
7# Create/edit skill config file for persistent settings
8# ~/.openclaw/skills/larry.yaml
9larry:
10  session_history: 50          # number of past interactions to retain in context
11  default_format: "checklist"  # default output format for task sequences
12  verbose_steps: true          # show step explanations during task sequences
13
14# Apply changes
15clawhub reload

Open OpenClaw chat and run a test prompt to confirm Larry is active and responding correctly. Start with a simple task that has a clear sequence of steps — something you would normally do manually in a checklist. This gives you a clean baseline before building more complex automation. Larry activates when OpenClaw routes productivity and task management intent to it. You can invoke it explicitly by name ('Larry, do X') or implicitly through prompts that describe task sequence needs. Explicit invocation is more reliable for first-time testing. A good test prompt is a short checklist walkthrough: ask Larry to prompt you through 3-4 steps of something you do regularly, capture your responses, and summarize the result. This exercises the core task automation loop without requiring any external integrations. If Larry does not respond or OpenClaw routes the prompt elsewhere, try prefixing your prompt with 'Use Larry to...' — this makes the skill routing intent unambiguous.

Once basic task sequences work, define a reusable workflow that Larry can run on demand without you re-describing the full sequence each time. This is the productivity multiplier: instead of reconstructing a complex workflow from scratch each morning, you invoke it with a short phrase and Larry executes the full sequence. You can define reusable workflows in two ways: either describe them conversationally to Larry and ask it to remember them for future invocations within the session, or define them as named templates in the larry.yaml skill configuration file for persistence across sessions and OpenClaw restarts. Named templates in the configuration file are more reliable for truly recurring workflows. Each template has a name, a description, and a sequence of steps — you then invoke it from OpenClaw chat by name rather than re-specifying all the steps. RapidDev can help teams define and maintain Larry workflow templates for common team processes — especially useful when standardizing workflows across multiple team members who share an OpenClaw setup.

1# Define named workflow templates in ~/.openclaw/skills/larry.yaml
2larry:
3  workflows:
4    weekly-review:
5      description: "End-of-week reflection and planning sequence"
6      steps:
7        - prompt: "Review completed tasks from this week. How many were finished vs planned?"
8        - prompt: "Name 3 wins from this week worth celebrating."
9        - prompt: "What 2 things went less well, and what will you do differently?"
10        - prompt: "Set your top 3 priorities for next week."
11      output_format: "summary"
12    daily-standup:
13      description: "Morning standup checklist"
14      steps:
15        - prompt: "What did you complete yesterday?"
16        - prompt: "What are you working on today?"
17        - prompt: "Any blockers or risks?"
18      output_format: "standup"

Larry's value compounds when combined with other ClawHub skills. A workflow that Larry orchestrates can trigger actions in other tools — save a weekly review summary to Supermemory for future reference, send a standup message to a Feishu or Discord channel, or update a data record via ByteRover based on task completion data. Combining skills in OpenClaw works through prompt chaining: you describe a multi-tool workflow in a single OpenClaw prompt, and OpenClaw routes each component to the appropriate skill. Larry handles the interactive task sequencing portion, and other skills handle the output actions. For example, a morning workflow might use Larry to step through your daily standup questions, then automatically format and send the standup summary to your team Feishu channel via feishu-bridge, and save the summary to Supermemory for end-of-week review. This entire sequence runs from one multi-step OpenClaw prompt.