AI Growth Engine
ReviewAudited by ClawScan on May 10, 2026.
Overview
No executable malware or credential use is shown, but the skill asks the agent to keep permanent growth memory and perform recurring self-review without clear boundaries.
Install only if you want the agent to maintain an ongoing growth journal. Before use, set explicit workspace paths, retention limits, deletion rules, and approval requirements for any heartbeat or memory writes, especially if tasks involve medical, student, business, or other sensitive information.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Private task details, sensitive work context, or mistaken conclusions could be retained and reused in later interactions.
The skill tells the agent to persist results into a growth log and update a permanent internal model, but the artifacts do not define retention, data minimization, workspace boundaries, or user review for what is stored.
L — Learn 记录: 将验证结果写入成长日志，更新内部模型（永久记忆）
Require explicit user opt-in before writing memories, restrict logs to a chosen workspace, avoid sensitive personal/medical/student/business data by default, and provide deletion and retention controls.
If the host honors these heartbeat instructions, the agent may keep processing and updating growth state even when the user did not explicitly request that run.
This describes recurring heartbeat-driven activity that updates summaries, trends, recommendations, and reports without stating clear user confirmation, stop conditions, or scope limits.
## 每日 - 运行一轮RAPVL循环总结 - 更新成长历史趋势图 - 检查失败模式聚合，更新推荐策略 - 每10轮生成进化报告
Make heartbeat behavior opt-in, require confirmation before scheduled writes or updates, and document how users can pause, inspect, or delete generated state.
Users may assume there is a reviewed implementation behind the examples, but only the instruction documents were supplied here.
The reviewed package contains no code files, so the documented Python implementation is not available for review; any external implementation fetched separately would be outside this artifact set.
from skills.ai_growth_engine import AIGrowthEngine
Treat the Python API examples as conceptual unless the implementation is separately provided and reviewed; avoid installing unreviewed code from outside the supplied artifact set.