Security audit
Security checks for vulnerabilities and agentic risk
This is a Markdown-only planning skill that encourages structured plans before multi-step work, with no hidden execution, persistence, or credential handling found.
Install only if you want a broadly activated planning aid. For sensitive work such as production migrations, user data processing, moderation systems, or deployments, keep normal approval, privacy, backup, and access-control procedures in place; the skill provides planning structure, not a substitute for authorization or review.
Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.
## Task
[Clear statement of what needs to be done]
## Plan
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
- POST /api/posts/{id}/comments - create comment
- GET /api/posts/{id}/comments - list comments (with threading)
- PATCH /api/comments/{id} - edit own comment
- DELETE /api/comments/{id} - soft delete
**Verification:**
- All endpoints return correct JSON
- Threading structure properly nested
The description uses very broad activation language such as 'use for any task with multiple steps' and lists major categories like coding, writing, analysis, and problem-solving. In an agent environment, this can cause the skill to activate for a large share of normal requests, increasing prompt surface area and the chance that its instructions override or interfere with more appropriate task-specific behavior.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- New comments start as "pending"
- Admin dashboard for moderation
- Email notifications for new comments
- Auto-approve trusted users
**Verification:**
- Pending comments don't appear publicly
- Moderation actions work
This markdown example instructs migrating 100,000 user records between platforms, which is a behavior that could materially affect user data and privacy. While the plan includes validation and rollback, it does not explicitly warn readers about sensitive-data handling, access controls, or privacy implications of processing user records.