Ollama Load Balancer
v1.0.4Ollama load balancer for Llama, Qwen, DeepSeek, and Mistral inference across multiple machines. Load balancing with auto-discovery via mDNS, health checks, q...
⭐ 0· 155·2 current·2 all-time
byTwin Geeks@twinsgeeks
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (Ollama load balancer) matches the runtime instructions: auto-discovery, health checks, routing, and admin HTTP endpoints. Required binaries (curl/wget) and optional python/pip/sqlite3 are appropriate for a Python-based local service.
Instruction Scope
SKILL.md instructs the agent to pip install ollama-herd and run local commands (herd / herd-node) and to call local HTTP endpoints on localhost:11435. It does not instruct reading unrelated system files or exfiltrating secrets. It does include administrative endpoints (pull/delete models) — expected for a load-balancer but potentially powerful if misused.
Install Mechanism
There is no registry install spec; the README tells the user to pip install ollama-herd from PyPI. pip installs are common but will execute third-party code on the host — moderate risk. The SKILL.md points to a PyPI project and GitHub repo (traceable), which is better than an arbitrary download, but users should review the package/source before installing.
Credentials
The skill declares no required environment variables or credentials. The use of FLEET_MAX_RETRIES and runtime settings is plausible for a load balancer; no unexplained secret access is requested.
Persistence & Privilege
always:false (normal). The instructions create and use local config paths (~/.fleet-manager/...), run long-lived processes, and expose an admin HTTP interface. This is consistent with the stated functionality but means the package will persist on disk and run services — run with appropriate isolation and review.
Assessment
This skill appears internally consistent for running a local Ollama load balancer, but it relies on you pip installing a third-party package and running local servers that manage model downloads and expose admin HTTP endpoints. Before installing: (1) review the ollama-herd PyPI package and GitHub repo to ensure the code matches expectations, (2) prefer installing in a sandbox/VM or isolated network, (3) disable or review auto-pull behavior to avoid unexpected large downloads, and (4) restrict access to the daemon's HTTP port (localhost-only or firewall) to prevent unauthorized remote control.Like a lobster shell, security has layers — review code before you run it.
auto-discoveryvk9713vhv3ymzcany9pxnhrm4b1840xmrdeepseekvk9713vhv3ymzcany9pxnhrm4b1840xmrdistributed-inferencevk97a724b4c1pbbktmnmhdw3mg183fr3yfailovervk9713vhv3ymzcany9pxnhrm4b1840xmrfleet-routingvk972ksse6jysj84ye4g8yms6d183xzcfhealth-checkvk9713vhv3ymzcany9pxnhrm4b1840xmrhigh-availabilityvk97a724b4c1pbbktmnmhdw3mg183fr3yinference-routingvk972ksse6jysj84ye4g8yms6d183xzcflatestvk9776zcn9gp9jphmrsz1mxnnxs8445cellamavk9713vhv3ymzcany9pxnhrm4b1840xmrload-balancervk9713vhv3ymzcany9pxnhrm4b1840xmrmistralvk972ksse6jysj84ye4g8yms6d183xzcfmonitoringvk9713vhv3ymzcany9pxnhrm4b1840xmrollamavk9713vhv3ymzcany9pxnhrm4b1840xmrqwenvk9713vhv3ymzcany9pxnhrm4b1840xmrself-hostedvk9713vhv3ymzcany9pxnhrm4b1840xmr
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
scales Clawdis
OSmacOS · Linux · Windows
Any bincurl, wget