sillytavern-cards-cn

WarnAudited by ClawScan on May 18, 2026.

Overview

This roleplay skill is coherent, but it lets imported character cards persistently rewrite the agent’s identity and memory, so users should review it carefully before installing.

Install only if you are comfortable with character cards changing OpenClaw’s persistent identity and memory. Prefer trusted cards, inspect imported prompts and knowledge-book entries, and know how to restore SOUL.md and clean MEMORY.md before using the agent for normal or sensitive work.

Findings (4)

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.

What this means

A downloaded character card could change how the agent behaves beyond a single conversation, including overriding normal assistant behavior.

Why it was flagged

The skill takes instructions embedded in imported character cards and writes them into the agent’s core identity file, making untrusted role-card prompt text authoritative for future behavior.

Skill content
从 `~/.openclaw/characters/<名字>.json` 读取角色 JSON,然后**覆写** `~/.openclaw/SOUL.md` ... `## 系统指令` `{{system_prompt}}` `{{post_history_instructions}}`
Recommendation

Only import character cards from sources you trust, inspect card prompts before activating play/soul modes, and keep a recoverable backup of SOUL.md.

What this means

Using link import sends requests to third-party card sites or arbitrary provided URLs and stores the resulting content locally.

Why it was flagged

The documented import flow downloads user-supplied URLs and writes parsed character data into local OpenClaw files. This is expected for the skill, but users should notice the network and file-write behavior.

Skill content
curl -sL "<url>" -o /tmp/card-download.png
node {baseDir}/extract-card.js /tmp/card-download.png > ~/.openclaw/characters/<角色名>.json
Recommendation

Use explicit, trusted card URLs and avoid importing unknown files or links without reviewing the resulting character data.

What this means

Imported card content or personal roleplay details may remain in memory after the character is stopped and could affect later agent responses.

Why it was flagged

The skill persists both card-provided knowledge-book content and relationship memories in the agent’s shared memory, where they can influence later sessions and retain personal conversation details.

Skill content
把知识书条目写入 `MEMORY.md` ... 聊天过程中,技能会把关系记忆保存到 `MEMORY.md` ... `/character stop` 会从备份恢复原来的 `SOUL.md`。关系记忆保留在 `MEMORY.md` 里
Recommendation

Review and clean MEMORY.md when removing or changing characters, and avoid importing cards whose knowledge-book or prompt content you have not inspected.

What this means

The agent may continue acting as the imported character across sessions, and in soul mode that persona can coexist with normal tool and skill use.

Why it was flagged

The skill deliberately creates persistent agent-persona modes, including one where the character personality remains active while all other skills continue to work.

Skill content
| **扮演** | `/character play <名字>` | 完全覆写为角色 | 禁用——纯角色扮演 | 永久,直到 `/character stop` | ... | **灵魂** | `/character soul <名字>` | 融合——角色人格 + 正常能力 | 所有技能正常使用 | 永久,直到 `/character unsoul` |
Recommendation

Use play/soul modes only when you want persistent behavior, and confirm that `/character stop` or `/character unsoul` has restored the agent before sensitive tasks.