Security audit
Security checks for vulnerabilities and agentic risk
This is a local article workflow guide, but it includes unguarded commands that can permanently change or delete article records.
Review this skill before installing. Use it only for the intended local article review platform, and require explicit user confirmation before any POST, PUT, publish-status update, publish-log write, or DELETE command. Treat the DELETE example as dangerous unless the platform provides backup, soft delete, or recovery.
The documentation presents article deletion as a simple one-line command with no warning, confirmation, authorization check, or rollback guidance. In an agent setting, this can normalize destructive behavior and make accidental or premature deletion of content records much more likely.
The trigger list includes generic terms such as 'review', '审稿', and '发布文章', which can match many ordinary conversations and cause the skill to activate outside its intended context. Over-broad activation increases the chance that the agent will expose internal API guidance or suggest state-changing actions during unrelated tasks.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -s http://localhost:3100/api/health
The manifest describes the skill as a complete guide for listing, reading, creating, reviewing, status updates, publishing sync, and review workflow. It does not mention destructive record removal, yet the file documents a DELETE endpoint for removing articles. Deletion is materially broader than the stated review-and-publish lifecycle.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# 拉取 approved 和 publishing 状态
curl -s "http://localhost:3100/api/articles?status=approved"
curl -s "http://localhost:3100/api/articles?status=publishing"
The instructions direct an agent to change article workflow state and append publish logs without any explicit confirmation, authorization check, or warning that these actions mutate system records and may trigger downstream publication behavior. In an agent setting, this increases the risk of unintended state changes, premature publication, or falsified operational history if the skill is invoked in the wrong context or by an over-trusting automation.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -s "http://localhost:3100/api/articles/{id}/content"
# 或读本地文件
cat "{project_path}/draft.md"
The guidance includes state-changing PUT requests that modify review metadata and article status, but it provides no warning, confirmation step, or change-control guidance before performing those actions. In an agent skill context, this can cause unintended workflow transitions or data integrity issues if an agent follows the instructions automatically or with insufficient user awareness.
The skill includes ready-to-run POST and PUT calls that modify a local service's article records, but it does not clearly warn the user that following the instructions will create or change persistent data. In an agent context, operational markdown can be treated as executable guidance, so this increases the risk of unintended state changes to a local content-management workflow.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -s -X POST "http://localhost:3100/api/articles" \
-H "Content-Type: application/json" \
-d '{
"title": "文章标题",
A language-only instruction set can violate organizational language/locale policy when it implicitly forces one language without offering alternatives or documenting that the skill is region-specific. This file provides all operational guidance solely in Chinese and does not state that the locale restriction is intentional or optional.
All user-facing instructions in this skill are presented in Chinese, and the file does not offer an alternative language or indicate that Chinese is a required or user-selected locale. This may violate language-choice policy when no opt-in or justification is provided.