Ai Prompt Optimizer
v1.0.0AI 提示词自动优化器。基于 A/B 测试和性能数据，自动优化提示词，提升输出质量和降低 Token 消耗。Triggers: prompt optimizer, improve prompt, prompt engineering, A/B testing, prompt optimization.
⭐ 0· 12·0 current·0 all-time
by@sky-lv
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (prompt optimizer, A/B testing) align with the SKILL.md content. All described capabilities (analysis, variant generation, optimization, A/B testing, template library) are coherent with a pure guidance/tooling skill.
Instruction Scope
SKILL.md is high-level and prescriptive about running A/B tests and collecting metrics (traffic splits, user satisfaction, token usage) but does not specify how to implement measurement, traffic routing, or telemetry. That vagueness gives the agent discretion to request integrations or data from the environment, so the runtime behavior depends on how the agent implements the guidance.
Install Mechanism
No install spec and no code files — lowest-risk profile. Nothing will be downloaded or written by the skill itself.
Credentials
The skill declares no required environment variables, credentials, or config paths, and the instructions do not reference secrets or unrelated system resources.
Persistence & Privilege
always is false and the skill does not request persistent presence or modify other skills/configs. Autonomous invocation is allowed (platform default) but not combined with other risky requests.
Assessment
This skill appears coherent and low-risk because it's an instruction-only prompt optimization guide that asks for no installs or credentials. Two practical cautions: (1) The A/B testing and metrics sections are conceptual — to actually run tests you (or the agent) will need ways to route traffic, record outputs, and measure user satisfaction; that typically requires model API keys, logging/analytics, or access to user interaction data that are not declared here. (2) Before using this skill to run large-scale experiments, decide where test data and outputs will be stored, whether user data will be included, and which credentials/integrations the agent may request; restrict or review any credentials the agent asks for and confirm no sensitive user data will be sent to external services. If you want to install, ask the author or integrator how A/B testing is implemented in your environment and which external integrations (model APIs, analytics) the agent will use.Like a lobster shell, security has layers — review code before you run it.
latestvk97bxsfdd6be27z9w0807jdckx84m0cr
License
MIT-0
Free to use, modify, and redistribute. No attribution required.