agenttrace
VerifiedLocal TUI for reviewing AI coding agent sessions, costs, and slowdowns.
What is agenttrace?
AgentTrace is an open-source Go application that delivers a terminal user interface and report generator focused on AI coding agent activity. It ingests session files from supported tools and presents aggregated metrics on usage and efficiency.
The tool processes logs locally to calculate costs, identify long pauses or retry patterns, and rank sessions by health or duration. Reports can be exported for further review while source prompts and code remain on the device.
It serves developers and teams that rely on AI coding agents and want structured insight into their collective output and bottlenecks.
Capabilities
What you can build with agenttrace
Cost and Token Tracking
Compare total spend, input/output tokens, and model usage across many past sessions from different agent sources.
Slow-Run Diagnosis
Surface long gaps, hanging calls, large parameters, and context pressure that explain extended task durations.
Session Ranking and Search
Order runs by cost, health score, or failures and locate specific sessions through metadata search.
Install agenttrace
curl -sL https://raw.githubusercontent.com/luoyuctl/agenttrace/master/install.sh | shcurl -sL https://raw.githubusercontent.com/luoyuctl/agenttrace/master/install.sh | sh- 1Download and run the install script or use your package manager to add the binary.
- 2Execute agenttrace in a terminal to open the interactive TUI against detected logs.
- 3Run agenttrace --doctor to verify log locations and cache status.
- 4Add flags such as --overview -f json or -f html to produce machine-readable or web reports.
- 5Use --search with keywords to filter sessions without leaving the local environment.
agenttrace: pros & cons
Pros
- +Fully local operation protects sensitive prompts and logs
- +Broad support for logs from many popular coding agents
- +Combines cost summaries with concrete diagnostic signals
- +Lightweight terminal experience with export options
Cons
- –Requires existing agent log files in recognized formats
- –Terminal UI may feel limited for users preferring graphical dashboards
- –Windows installation path differs from Unix defaults
Frequently asked questions
No. All parsing and reporting occur on the local machine.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other coding & dev options worth comparing.
