GlassPipe is an open-source tool designed to record and visualize the activity of AI agents, specifically capturing every large language model (LLM) call made during an agent's execution. By applying a single decorator to a Python function, users can automatically trace all LLM interactions, generating a detailed visual timeline of each run. This approach allows developers and students to inspect, debug, and understand the internal workings of their AI agents without altering their existing codebase beyond the addition of the decorator.
The tool provides a local dashboard that displays each agent run as an interactive timeline, including details such as LLM model used, input and output data, duration, and associated costs. Users can click on individual spans within the timeline to inspect specific calls in detail. GlassPipe also enables sharing of these traces via public links, allowing others to view the exact same timeline through a web interface. Before sharing, users are presented with a pre-share preview that scans for common secrets such as API keys, tokens, and emails, and allows for manual redaction to ensure no sensitive information is uploaded. Publicly shared traces expire after 30 days, and users receive a delete token to remove a trace early if desired.
GlassPipe is built for synchronous Python environments and currently supports tracing LLM calls made through the OpenAI and Anthropic SDKs. The tool is targeted at indie developers, students, and anyone seeking quick observability and sharing of agent traces without the need for account creation or API keys. It is not intended for production monitoring or enterprise use cases that require features such as team workspaces, alerting, or compliance certifications.
dev service. 10 and above. The project is maintained by Jonathan and is open source.
GlassPipe is a LLM eval & observability product. It focuses on debugging and visualizing the sequence of LLM calls made by AI agents for transparency and troubleshooting. It is built as an open-source project for AI developers and researchers. GlassPipe is open source under the MIT license. It runs on the web and the command line, and it can be self-hosted.
GlassPipe first shipped in 2026. Development happens publicly on GitHub with 107 commits in the last 90 days. PulseGate's similarity index finds few close equivalents — GlassPipe occupies a relatively distinct niche. Key capabilities include LLM call tracing, timeline visualization, and shareable traces.
Latest indexed changes and source events
glasspipe.dev discovered by the PulseGate indexer
Other apps tracked under the same category.
