Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Green Vault — AI Agent 安全与绿色运维顾问
v1.0.0AI Agent 安全与绿色运维顾问。合并 EcoCompute（GPU 能耗优化）与 OpenClaw/Bagman（安全密钥管理）， 提供 LLM 推理部署的能效分析、密钥安全管理、泄露防护和注入防御一体化方案。 Use when handling GPU energy optimization for LL...
⭐ 0· 51·0 current·0 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 optimization + key management) align with requested binaries (nvidia-smi for NVML power readings, python for scripts, op for 1Password). No unrelated credentials or surprising tools are requested.
Instruction Scope
SKILL.md and references limit actions to energy benchmarking, config recommendations, and secret-safe access patterns using the 1Password CLI. Example code shows subprocess calls to 'op' and output-sanitization, which are expected for on-demand secret retrieval and leak prevention. The files do not instruct reading unrelated system config or exfiltrating data to unknown external endpoints.
Install Mechanism
Instruction-only skill with no install/spec downloads. This is the lowest-risk install pattern and is appropriate for documentation / runbook style skills.
Credentials
No environment variables or additional credentials are declared. Requiring the 'op' CLI is proportionate to the described 1Password-based secret workflow. Users should note that runtime use of 'op' will access whatever 1Password account/session is available to the agent, so service-account least-privilege is recommended.
Persistence & Privilege
Skill is not always-enabled and does not request elevated platform privileges or modify other skills. It recommends ephemeral session keys and explicitly forbids persisting raw private keys, matching a low-privilege operational model.
Scan Findings in Context
[prompt-injection:ignore-previous-instructions] expected: The SKILL.md and references include prompt-injection test patterns and defenses (e.g., validating input for 'ignore previous instructions'). Detection of that pattern is expected because the skill documents injection attacks and includes test cases/rules to block them.
Assessment
This skill appears to do what it says: energy profiling and secure runtime secret access via the 1Password CLI. Before installing or running it, consider: (1) the agent will call nvidia-smi and may execute python subprocesses (including 'op') — ensure you only grant it access to a dedicated 1Password service account or vault with minimal permissions; (2) confirm you trust the skill source and review the full SKILL.md (the manifest here shows a truncated code snippet) to ensure nothing unexpected is omitted; (3) since the skill fetches secrets at runtime, verify your operator policies, auditing, and revocation processes for any session keys used. If you want extra assurance, request the full, untruncated SKILL.md and a provenance statement (who maintains the skill and where its empirical data originates).references/prompt-injection-defense.md:81
Prompt-injection style instruction pattern detected.
About static analysis
These patterns were detected by automated regex scanning. They may be normal for skills that integrate with external APIs. Check the VirusTotal and OpenClaw results above for context-aware analysis.Like a lobster shell, security has layers — review code before you run it.
1passwordvk973gfdpwbpvrq6paxk52ge8bh83fjm9ai-agentvk973gfdpwbpvrq6paxk52ge8bh83fjm9energy-efficiencyvk973gfdpwbpvrq6paxk52ge8bh83fjm9erc-4337vk973gfdpwbpvrq6paxk52ge8bh83fjm9gpuvk973gfdpwbpvrq6paxk52ge8bh83fjm9key-managementvk973gfdpwbpvrq6paxk52ge8bh83fjm9latestvk973gfdpwbpvrq6paxk52ge8bh83fjm9llm-inferencevk973gfdpwbpvrq6paxk52ge8bh83fjm9securityvk973gfdpwbpvrq6paxk52ge8bh83fjm9
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🦞🔐 Clawdis
Binsnvidia-smi, python, op