prompt-Compres龙虾必备：提示词大师@博采众长-自我进化🐂越用越省Token
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The skill is mostly what it claims (a prompt optimizer) but includes an automatic self-learning subsystem that contacts obfuscated remote servers and will attempt background cloud sync of local learning records—behavior that is surprising and potentially exposes prompt content.
This skill implements a local optimizer but also includes an automatic self-learning system that uploads (allegedly desensitized) learning data and prompt examples to remote servers whose endpoints and keys are obfuscated. If you plan to use it: - Read SKILL.md and the scripts before running. The cloud-sync behavior is declared, but the code hides the API endpoints/keys (obfuscated) and derives keys from your machine identity. - Do NOT use this skill with sensitive prompts or secrets you wouldn't want transmitted remotely (credentials, PII, proprietary text). - If you want local-only behavior, disable the self-learning/sync by removing or renaming the scripts folder or the files scripts/prompt_learning.py and scripts/self_grow.py (prompt_master checks existence and will skip recording if the learning script is absent). Alternatively edit prompt_master.py to remove or comment out the call to record_to_learning. - If you must allow sync, prefer running in a network-restricted sandbox or VM so you can monitor/limit outbound connections, and verify the remote endpoints and privacy policy with the author. The obfuscation and direct IP/address in code reduce transparency — ask the publisher for a clear privacy/security statement and the real server hostname and data retention policies before enabling sync. Given the obfuscation and automatic uploads, treat this skill as potentially privacy-risky until you confirm what data is sent and to whom.
SkillSpector
SkillSpector findings are pending for this release.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
No VirusTotal findings