Use this skill when the user asks to export audit logs, find audit log location, view command history, 导出日志, 查看日志, 日志路径, 操作记录, 调用记录, 命令历史. Do NOT use for wal...
Security Analysis
high confidenceThe skill's stated purpose (tell the user where Onchain OS writes audit logs and describe their format) is plausible and low-risk, but the runtime instructions reference an environment variable ($ONCHAINOS_HOME) that is not declared in the skill metadata — a small but meaningful incoherence.
Name and description align with the instructions: the skill only documents the audit log path and format for Onchain OS. No credentials, binaries, or installs are requested, which is proportional. However, the SKILL.md refers to an environment variable ($ONCHAINOS_HOME) for an alternate path while the skill metadata declares no required env vars — a mismatch between documentation and declared requirements.
Instructions are narrowly scoped (provide path, format, fields, rotation) and explicitly forbid reading/displaying file contents. The concerning part is that SKILL.md instructs using $ONCHAINOS_HOME if set — that implies the agent may check environment variables at runtime even though none are declared. The instructions do not explain how to obtain values for placeholders (os/arch/cli_version) and could cause the agent to probe environment/system state.
No install spec and no code files — instruction-only. This is the lowest-risk installation model and consistent with the skill's simple documentation purpose.
The skill requests no credentials or config paths (appropriate for its stated purpose). The only discrepancy is the undocumented reference to $ONCHAINOS_HOME in SKILL.md; reading that single env var could be reasonable, but it should have been declared in requires.env to be transparent.
always is false and there is no install or persistent behavior. The skill does not request persistent presence or elevated privileges.
Guidance
This skill appears to only tell you where Onchain OS keeps its audit log and what the file contains; it does not request credentials and explicitly forbids reading log contents. However, the instructions reference an environment variable ($ONCHAINOS_HOME) that the skill metadata does not declare — ask the publisher why that env var is referenced and whether the agent will check it at runtime. Also verify the skill's provenance (owner ID here is unfamiliar) before trusting it in a production agent. Avoid pasting or asking the agent to read actual log file contents unless you trust the skill and understand the sensitivity of the logs.
Latest Release
v2.4.0
- Version update to 2.4.0 (from 2.2.10) in metadata. - No other changes detected.
More by @ok-james-01
Published by @ok-james-01 on ClawHub
