Releases·Tool Release
Dev launches LLMxRay, a 100% local observability engine for tracking tool calls, latency, and token usage in local LLM stacks
A developer launched LLMxRay, a 100% local observability and analytics engine for self-hosted LLM inference and agentic workflows. It proxies traffic to engines like Ollama and exposes tool-call tracing, latency breakdowns, and token metrics without sending data externally. It targets the gap left by managed API providers' built-in dashboards, where local setups lack visibility into failures and usage patterns.
1 engagement1 sourceSun, Oct 4, 2026, 06:52 AM
AI Overview
- LLMxRay sits in front of local inference engines like Ollama, capturing telemetry without external data transmission or added latency
- The tool intercepts agent tool/function calls, inspecting payloads, call frequencies, error rates, and multi-step loops where agents stall
- Latency profiling breaks requests into Time-To-First-Token, prefill duration, and decode speed, alongside token throughput and per-model utilization
- LLMxRay launches via one command (npx llmxray) with a local dashboard at localhost:3000 and no cloud dependency
- Developer LogneBudo released the project on GitHub (github.com/LogneBudo/llmxray) seeking feedback from teams running local LLM workloads
Related
Products·building
Tracewell: Open-source observability dashboard for AI agents
Tracewell is an early open-source observability tool that lets developers add a tiny Python SDK to their AI agent applications to inspect traces locally in a dashboard. It captures prompts, responses, latency, token usage, estimated cost, success/error status, tool calls, and rule-based failure detection, helping developers debug and monitor agent performance.
Tracewell is an early open-source observability tool that lets developers add a tiny Python SDK to their AI agent applications to inspect traces locally in a dashboard. It captures prompts, responses, latency, token usage, estimated cost, success/error status, tool calls, and rule-based failure detection, helping developers debug and monitor agent performance.
1 engagement1 sourceredditTue, Jul 28, 2026, 12:26 PM
Releases·Tool Release
LLMPeek: open-source tool for local LLM token monitoring in development
A developer released LLMPeek, an open-source tool that lets developers monitor LLM token usage, latency, and cost in real-time during app development. It works by adding one import to Node apps or running a local proxy, capturing all requests to OpenAI, Anthropic, or any OpenAI-compatible API without sending data to the cloud.
A developer released LLMPeek, an open-source tool that lets developers monitor LLM token usage, latency, and cost in real-time during app development. It works by adding one import to Node apps or running a local proxy, capturing all requests to OpenAI, Anthropic, or any OpenAI-compatible API without sending data to the cloud.
1 engagement1 sourceredditTue, Jul 14, 2026, 10:12 PM
Products·launched
Local terminal observability dashboard for Claude Code and Codex
Aether provides a live terminal dashboard that visualizes native local traces from Claude Code and Codex, showing context growth, cost, latency, tools, compactions, code changes, agents, and per-turn input work. It requires no SDK, API key, cloud backend, or additional model calls, solving the problem of understanding and debugging LLM agent behavior locally for developers using these tools.
Aether provides a live terminal dashboard that visualizes native local traces from Claude Code and Codex, showing context growth, cost, latency, tools, compactions, code changes, agents, and per-turn input work. It requires no SDK, API key, cloud backend, or additional model calls, solving the problem of understanding and debugging LLM agent behavior locally for developers using these tools.
1 engagement1 sourceredditFri, Jul 17, 2026, 07:39 PM
Releases·Tool Release
Blazo: open-source local, real-time observability for AI agents (MVP on GitHub)
Blazo, an open-source observability product for AI agents, is available on GitHub as an MVP. It ingests agent telemetry — runs, traces, tool calls, LLM calls, errors, latency, tokens — and shows execution live in a local UI with no account required. It targets developers debugging why agents are slow, stuck, or failing, and is explicitly not an agent builder, playground, or evaluation platform.
Blazo, an open-source observability product for AI agents, is available on GitHub as an MVP. It ingests agent telemetry — runs, traces, tool calls, LLM calls, errors, latency, tokens — and shows execution live in a local UI with no account required. It targets developers debugging why agents are slow, stuck, or failing, and is explicitly not an agent builder, playground, or evaluation platform.
45 engagement1 sourcegithubSun, Sep 27, 2026, 01:55 PM
Products·building
LLM Inference Dashboard — local resource/metrics dashboard for multiple inference engines
A lightweight local dashboard that provides richer logging and metrics for LLM inference than the default endpoint APIs expose. It aggregates resources across multiple engines (llama, custom CUDA, unsloth, etc.) via a collector, scaling to network API endpoints. Aimed at local LLM users who run more than one engine.
A lightweight local dashboard that provides richer logging and metrics for LLM inference than the default endpoint APIs expose. It aggregates resources across multiple engines (llama, custom CUDA, unsloth, etc.) via a collector, scaling to network API endpoints. Aimed at local LLM users who run more than one engine.
16 engagement1 sourceredditSun, Oct 4, 2026, 06:00 PM