Baoyu Wechat Summary

WarnAudited by ClawScan on May 14, 2026.

Overview

The skill is coherent for summarizing WeChat chats, but it needs unsandboxed access to local WeChat data and keeps long-term per-user chat profiles.

Install only if you are comfortable granting the agent unsandboxed wx-cli access to your local WeChat data. Manually configure EXTEND.md if possible, verify the wx-cli source/version, run it only for intended groups and time ranges, and regularly review or delete the generated history and profile files.

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

wx-cli commands would run with broad local permissions while handling WeChat data, so a mistake or unsafe command could affect or read more than the user intended.

Why it was flagged

The workflow asks the agent to bypass normal sandbox containment for all wx-cli operations, not just a narrow one-off command.

Skill content
Every `wx` command in this skill needs to run with `dangerouslyDisableSandbox: true` from the start — don't waste a sandbox attempt first.
Recommendation

Use only if you trust the local wx-cli binary and are comfortable with unsandboxed execution; prefer explicit user approval before each wx command.

What this means

The agent may access sensitive WeChat account data, local chat databases, and recent session information beyond the specific chat the user wants summarized.

Why it was flagged

The skill uses local WeChat account/session configuration and can inspect recent private or group sessions for setup, which is broader than summarizing one selected group.

Skill content
wx-cli reads from `~/.wx-cli/` (config, cache, daemon socket) and from WeChat's data directory ... `wx sessions --json --limit 20` ... scan the sessions for any private/group thread
Recommendation

Create EXTEND.md manually with self_wxid/self_display when possible, review the wx-cli commands before running, and avoid using this on highly sensitive WeChat accounts.

What this means

Running the skill can create long-lived local dossiers about group members that may persist and be reused across future summaries.

Why it was flagged

The skill creates persistent per-user memory from private group chats, including quotes, events, role labels, and roast material, with append-only/no-cap retention.

Skill content
Profiles let the digest carry forward observations across many days ... `经典金句` **Append-only** Never delete. No cap. ... `毒舌语录库` **Append-only** Never delete. No cap.
Recommendation

Review and periodically delete the generated `wechat/` history and `profiles/` or `profiles-roast/` folders; avoid running on groups where participants would not expect persistent profiling.

What this means

Security depends partly on the installed wx-cli package and its local configuration.

Why it was flagged

The skill is instruction-only and depends on an external local binary rather than reviewed bundled code; this is disclosed and purpose-aligned, but users must trust that binary.

Skill content
底层依赖外部 [wx-cli](https://github.com/jackwener/wx-cli) 二进制(`wx` 命令),不打包脚本。
Recommendation

Install wx-cli only from a trusted source, pin or record the version you use, and update it deliberately.