In the OpenClaw ecosystem, tools that empower local AI assistants to document their work in real time are gaining traction. The Showboat pattern, a method for coding agents to create Markdown documents showcasing their code, has recently been extended with two new additions: Chartroom and datasette-showboat. These tools, originally developed by Simon Willison, align perfectly with OpenClaw’s vision of open-source, local-first AI automation, enabling agents to stream updates and embed visualizations seamlessly.
Chartroom serves as a CLI charting tool designed to integrate with Showboat, allowing agents to generate PNG images for charts that can be embedded directly into documentation. Meanwhile, datasette-showboat leverages Showboat’s remote publishing feature to incrementally push documents to a Datasette instance, facilitating real-time viewing of agent progress. For OpenClaw users, this means enhanced capabilities for monitoring and interacting with AI-driven workflows without relying on cloud dependencies.
How Datasette-Showboat Enhances OpenClaw’s Remote Monitoring
Showboat version 0.6.0 introduced a remote feature that operates via an environment variable, such as export SHOWBOAT_REMOTE_URL=https://www.example.com/submit?token=xyz. This setup causes every showboat init, note, exec, or image command to POST document fragments to a specified API endpoint, in addition to updating the local Markdown file. The API uses simple POST form variables or multipart form uploads for images, as detailed in the Showboat README.
Datasette-showboat is a Datasette plugin that adds endpoints for viewing documents at /-/showboat and receiving updates at /-/showboat/receive. To deploy it quickly, users can run commands like uvx --with datasette-showboat --prerelease=allow datasette showboat.db --create -s plugins.datasette-showboat.database showboat -s plugins.datasette-showboat.token secret123 --root --secret cookie-secret-123, then set the environment variable to export SHOWBOAT_REMOTE_URL="http://127.0.0.1:8001/-/showboat/receive?token=secret123". After running uvx showboat init demo.md "Showboat Feature Demo", documents become viewable in real time through the Datasette interface.
This functionality is particularly valuable in the OpenClaw context, as it allows agents like those in Claude Code for web to transmit documents to a server hosted on platforms such as Fly.io. By setting the SHOWBOAT_REMOTE_URL in the agent’s environment, users can monitor progress and provide feedback during active sessions, enhancing collaboration between humans and AI assistants. Tools like Rodney, a CLI browser automation tool built for Showboat, further complement this by enabling web interactions and screenshot captures that integrate with remote publishing.
Chartroom: Visualizing Data in OpenClaw Workflows
Chartroom addresses the need for visual data representation in agent documentation. As a wrapper around the matplotlib Python library, it supports chart types including bar, line, scatter, and histograms. For example, to create a bar chart, users can pipe CSV data: echo 'name,value
Alice,42
Bob,28
Charlie,35
Diana,51
Eve,19' | uvx chartroom bar --csv --title 'Sales by Person' --ylabel 'Sales'. This outputs a PNG image that can be embedded into Showboat documents.
Additional features include alt text generation with the -f alt flag, which produces descriptive text like “Sales by Person. Bar chart of value by name — Alice: 42, Bob: 28, Charlie: 35, Diana: 51, Eve: 19”, and support for Markdown or HTML output with -f markdown or -f html. Showboat version 0.5.0 added compatibility for Markdown images with alt text, aligning with Chartroom’s capabilities. Chartroom also offers various matplotlib styles, demonstrated in documents like demo/styles.md.
From an OpenClaw perspective, Chartroom enables local AI assistants to perform data analysis tasks—such as querying SQLite databases—and produce visual summaries within documentation. This reduces reliance on external charting services and keeps workflows contained within the local ecosystem.
Building Tools for the OpenClaw Ecosystem
The development of Chartroom exemplifies how tools can be crafted for agent use in OpenClaw. Starting with a click-app cookiecutter template, the process involved instructing a Claude Code for web session to build a Python CLI tool with subcommands for chart types, data input from CSV, TSV, JSON, or SQL queries, and integration with Showboat for demo documentation. The spec included maintaining a demo/README.md using Showboat, with each new chart type committed alongside tests and implementation.
This approach highlights the parallel agent workflows possible in OpenClaw, where multiple tools can be developed simultaneously, such as datasette-showboat alongside Showboat’s remote feature. The environment variable mechanism for remote streaming acts as a webhook extension, opening doors for future innovations in the ecosystem.
The Growing Showboat Family in OpenClaw
The Showboat family now includes Showboat itself, Rodney for browser automation, Chartroom for charting, and datasette-showboat for remote document streaming. These tools operate on loose conventions: any tool outputting image paths can be included in Showboat documents, and any text-outputting tool can integrate with Showboat. This flexibility encourages rapid tool development within the OpenClaw framework, supporting diverse automation needs.
For OpenClaw users, this ecosystem means enhanced documentation capabilities, real-time monitoring, and visual data integration—all while maintaining a local-first, open-source philosophy. As more tools adopt this pattern, the potential for innovative agent workflows continues to expand.
