Psych Companion
v0.1.0心理陪伴助手 - 情绪记录、压力评估、调节建议，危机时刻的专业转介支持
⭐ 0· 65·0 current·0 all-time
byhaidong@harrylabsj
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (daily emotional support, CBT-based suggestions, crisis referral) match the provided files: emotion analyzer, crisis detector, CBT references, and simple tests. No unrelated credentials, binaries, or external services are requested.
Instruction Scope
SKILL.md promises 'sensitive words will be cleaned immediately' and 'all processing is local, no persistent storage.' The shipped Python modules do not implement explicit redaction/sanitization of user text (detectors return matched keywords and generate responses) and there is no code that writes to persistent storage — so the persistence claim is accurate but the automatic 'cleaning' promise is not visibly enforced in the code. Recommend verifying how the agent runtime will handle logs/output and whether any wrapper will perform the claimed scrubbing.
Install Mechanism
Instruction-only install (no install spec). Code files are pure Python using only standard libraries (json, re, pathlib). No downloads, no external packages, no archive extraction — low install risk.
Credentials
No required environment variables, credentials, or config paths. The skill does not request unrelated secrets and its data access is limited to bundled reference JSON files.
Persistence & Privilege
always=false (not force‑installed) and no self-modifying or cross-skill configuration is present. The code does not persist user records; tests and scripts only read local reference files.
Assessment
This skill appears coherent and runs locally without requesting credentials or network access. Before installing/using it: 1) Review where your agent/runtime logs user input — the skill's code does not explicitly redact sensitive phrases even though SKILL.md claims immediate cleaning. 2) Run the included tests locally to confirm behavior and that no persistence or unexpected network activity occurs in your environment. 3) Understand it is NOT a replacement for professional care; verify the hotline number is appropriate for your region and ensure emergency instructions are correct. 4) If you need the claimed immediate privacy scrubbing, ask the author to show or add explicit redaction code (or add a preprocessing step in your agent) so that sensitive tokens are removed from outputs/logs.Like a lobster shell, security has layers — review code before you run it.
latestvk97dkw48vrchttwx8m9dnh3hd983hbf6
License
MIT-0
Free to use, modify, and redistribute. No attribution required.