1. Conduid
  2. Analytics
  3. LLM Hops
MCP server · Analytics

LLM Hops

by · github.com/YauhenBichel/llm-hops

See every hop of a request through your local LLM system: a tracing server, a live flow map, a waterfall per request, and an MCP server for Claude Code and Cursor. One static Go binary.

Unclaimed claude-codeclidebugginggogolanglatencyllmllm-gatewayllm-observabilityllmopslocal-llmmcpmcp-servermodel-context-protocolobservabilityollamarequest-tracingself-hostedtracingyserver
InstallView source
34Low

Scored 13 days ago · breakdown

About LLM Hops

LLM Hops is an MCP server published by YauhenBichel in the Analytics category: see every hop of a request through your local LLM system: a tracing server, a live flow map, a waterfall per request, and an MCP server for Claude Code and Cursor. One static Go binary.

The repository has 1 stars and 0 forks, with the last commit 21 days ago. Recent commit activity is the strongest signal that reported issues will get attention.

Install

From source
git clone https://github.com/YauhenBichel/llm-hops

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about LLM Hops

Powered by Claude · Grounded in docs

I know everything about LLM Hops. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

6 checks have not been run yet.

Releases

v0.2.4v0.2.4: the model timeline · 26 Sep 2026A band across the traces view: every model load from Ollama's journal as a mark, the model in the GPU as the band after it, with a legend and hover; `GET /api/v1/loads`. The traces list shows requests only by default;…
v0.2.3v0.2.3: model loads from Ollama's journal · 26 Sep 2026`serve -ollama-journal` (or `ollama_journal = true`): every model load in `journalctl -u ollama` becomes a trace of its own, with the model's name, the blob and the seconds; `ollama_since` reads the history first. Stats count loads and…
v0.2.2v0.2.2: the adapter learns trace ids; log-derived spans never overwrite a service's own · 26 Sep 2026yllm-gateway's log lines carry `trace_id` and the gateway posts its own spans; the adapter uses the same ids, so a trace fed both ways merges. Log-derived spans are inserted only where nothing exists (`AddIfAbsent`, `if_absent` on the…
v0.2.0v0.2.0: trustworthy on one box, local or cloud, integrations · 26 Sep 2026A configuration file (`llm-hops config`) with an optional token; `LLM_HOPS_TOKEN`. `/metrics` in Prometheus text; daily rollups per model so the stats reach past the span retention; `export` and `import`; `bench` (9,935 spans/s sustained…
v0.1.0v0.1.0: the first picture · 26 Sep 2026The first release. One static binary: spans in over HTTP, a SQLite store, traces and statistics out, a live stream, and the page (flow map with live requests, traces with a waterfall each, stats per hop and per model). An adapter follows…
Questions

About LLM Hops

How do I install LLM Hops?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is LLM Hops safe to use with an AI agent?

Its trust score is 34 out of 100 (low). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is LLM Hops still maintained?

Yes — the latest release is v0.2.4 (26 Sep 2026), and the last commit was 21 days ago. The repository has 1 stars.