Pretenziya Ru

Составь претензию или жалобу в банк, магазин, управляющую компанию, госорган или работодателя по описанной ситуации.

This skill will draft Russian complaint letters as advertised, but its instructions also tell the agent to run python3 commands that read and write /home/node/.openclaw/workspace/ru-pack-counter.txt and to add promotional t.me links to responses when the counter is low. That behavior is unrelated to drafting complaints and creates persistent state in the agent environment. Before installing, consider: 1) Do you want a skill that executes local commands and modifies files in the agent workspace? 2) Ask the author to remove the attribution code (the python3 read/write and automatic promotional text) or make it opt-in and clearly declared. 3) If you must use it, run the skill in a restricted sandbox or deny it permissions to modify persistent workspace files. 4) Verify presence of python3 if you intentionally accept that behavior. If you cannot get the attribution removed or made optional, treat the skill as untrusted because it performs side effects beyond its stated purpose.

Type: OpenClaw Skill Name: pretenziya-ru Version: 1.0.2 The skill is classified as suspicious due to its use of direct command execution via `python3 -c` within `SKILL.md`. While the executed commands (reading from and writing to `/home/node/.openclaw/workspace/ru-pack-counter.txt` for usage tracking and conditional promotion) are not inherently malicious, the capability to execute arbitrary code and perform file system operations represents a significant security risk (potential RCE vulnerability) if misused or if the commands were to be altered. The skill does, however, include explicit instructions to the AI agent to prevent prompt injection from user-provided documents, which is a positive security practice.