小五子 - 基因创新

通过跨域融合与动态调整，智能筛选并创新提示词逻辑，持续驱动基因演化和系统适应性提升。

Security Scan

The skill's name and high-level description match (prompt 'gene' innovation), but the provided SKILL.md is purely conceptual and vague, giving the agent broad, unspecified discretion — this mismatch in operational detail is a red flag.

Purpose & Capability

Name and description claim a prompt-evolution/innovation capability and there are no unrelated env vars, binaries, or installs requested — that part is coherent. However, the SKILL.md only describes high-level algorithmic concepts (cross-domain fusion, heuristic selection, dynamic mutation rates) without any concrete or proportional runtime requirements or boundaries.

Instruction Scope

The SKILL.md is conceptual and lacks concrete runtime instructions (no algorithms, no evaluation metrics, no data sources, no storage or API endpoints). It uses open-ended language that gives the agent broad discretion (e.g., 'simulate random logic', 'automatically adjust mutation rate', 'in long-running agents') — this can lead to the agent reaching for system state, files, or external services unless explicitly constrained.

Install Mechanism

No install spec and no code files are present, so nothing will be written to disk or downloaded by the skill itself. This minimizes installation risk.

Credentials

The skill declares no required environment variables, config paths, or credentials, which is proportionate for an instruction-only, conceptual prompt-evolution helper.

Persistence & Privilege

always is false and the skill is user-invocable. There is no request for persistent system presence or modification of other skills' configurations.

What to consider before installing

This skill is conceptually aligned with 'prompt evolution' but is ambiguous about what the agent should actually do at runtime. Before installing or enabling it for autonomous use, consider: 1) Ask the author for a concrete runtime spec: how are variants generated, how is fitness evaluated, what data sources and endpoints are used, and where are artifacts stored? 2) Require explicit boundaries: forbid access to system files, network endpoints, or credentials unless specifically needed and justified. 3) Start with manual (user-invoked) testing only — do not allow autonomous invocation until you review a concrete implementation or code. 4) If code is later provided, insist on a security review (esp. for any network calls, file I/O, or persistent storage). 5) Avoid running this skill on sensitive data or in environments with access to credentials until its behavior is well understood. These steps will reduce the risk that the skill's vague instructions cause unintended data access or external calls.

# 小五子 - 基因创新 (xiaowuzi-gene-innovate)

## 技能深度解析

该技能是小五子基因演化系统中的“变异与探索”核心。它不仅仅是简单的修改，而是致力于在现有的提示词基因库之外，挖掘具有突破性的逻辑结构。

## 核心算法能力

- **跨域思维融合**：通过模拟随机逻辑，将不同领域的逻辑范式引入当前的提示词设计中，产生意想不到的效果。

- **高维演化筛选**：利用启发式算法，在海量的变异分支中精准定位那些具备演化潜力的优秀基因片段。

- **动态变异率调整**：根据当前的演化停滞情况，自动调高或调低逻辑变异的剧烈程度，确保系统不陷入局部最优解。

## 部署价值

在长期运行的自动化代理任务中，该技能能有效防止系统逻辑僵化，为小五子提供持续的创新动力，确保在处理复杂、多变的现实任务时始终保持领先的灵活性和适应性。