EcoCompute — LLM Energy Efficiency Advisor
v2.5.0EcoLobster energy advisor: save 30-701% wasted GPU energy. RTX 5090 five-precision benchmarks (FP16/FP8/NF4/INT8-mixed/INT8-pure), 113+ measurements, dollar-...
⭐ 0· 826·1 current·2 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (GPU energy advisor, quantization/batch guidance) align with the declared requirements: it asks only for nvidia-smi and python (reasonable for NVML-based measurements and local GPU inspection) and includes extensive measurement/reference docs that support its claims.
Instruction Scope
SKILL.md is instruction-only and stays focused on advising/diagnosing inference energy use, giving code snippets, parameter validation, and persona rules. Two things to note: (1) the skill's persona and auto-activation wording is broad — it says it 'activates automatically when you discuss LLM deployment', which may cause it to trigger in many related conversations; (2) the AUDIT protocol asks to review user code/configuration (normal for an audit) — users should avoid pasting secrets. The skill does not instruct reading unrelated system files or exfiltrating data.
Install Mechanism
No install spec or executable payload is included (instruction-only). The MANUAL suggests cloning a GitHub repo or using an npx installer, but those are user-invoked steps — there is no automatic download/execute step embedded in the skill metadata.
Credentials
The skill requests no environment variables or credentials and declares no config path access. Requiring nvidia-smi/python is proportionate to GPU power monitoring and the measurement protocols described.
Persistence & Privilege
Flags: always:false and default autonomous invocation are normal. The skill does not request forced always-on presence or system-wide configuration changes in the provided files.
Assessment
This skill appears internally consistent with its stated purpose. Before installing: (1) Verify you trust the upstream repository (MANUAL recommends cloning from GitHub) — review that repo before running any install commands. (2) Be aware the skill may ask you to paste inference code/configs for an 'AUDIT' — avoid sharing secrets (API keys, private credentials) in those snippets. (3) If you allow the agent to run local diagnostics, confirm nvidia-smi/python are present and that you understand any commands it suggests to run. (4) If you prefer less intrusive behavior, disable automatic activation or limit the skill to user-invoked only. Finally, treat the measurement claims as empirical guidance — consider validating changes on a staging environment before applying to production.Like a lobster shell, security has layers — review code before you run it.
ai-mlvk978rzgpp4mx6a9ja481e045ah835rrjcarbon-footprintvk978rzgpp4mx6a9ja481e045ah835rrjenergy-efficiencyvk978rzgpp4mx6a9ja481e045ah835rrjgpuvk978rzgpp4mx6a9ja481e045ah835rrjgpu-optimizationvk97b2byqwdjbrvjf400cn3drf181atvdgreen-aivk978rzgpp4mx6a9ja481e045ah835rrjinferencevk978rzgpp4mx6a9ja481e045ah835rrjlatai-mlvk97enennavc4qcsp751djtr64d82m18qlatestvk978rzgpp4mx6a9ja481e045ah835rrjllmvk978rzgpp4mx6a9ja481e045ah835rrjmlvk972txnwx8eqmsa7kvyvn3jnrd835waxoptimizationvk978rzgpp4mx6a9ja481e045ah835rrjquantizationvk978rzgpp4mx6a9ja481e045ah835rrjsciencevk978rzgpp4mx6a9ja481e045ah835rrjsustainabilityvk978mhsnh7b1p22neqbpqe9qmx82q3k8sustainabilityestvk978rzgpp4mx6a9ja481e045ah835rrjutilityvk978rzgpp4mx6a9ja481e045ah835rrj
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
Binsnvidia-smi, python