Security audit
Security checks for vulnerabilities and agentic risk
The skill openly tells an agent to work independently on coding tasks, and although its trigger is broad, it does not include hidden code, persistence, credential harvesting, or data exfiltration.
Install this only if you want an agent to take a more autonomous approach to coding requests. Review changes before applying or committing them, especially in sensitive repositories, because the skill encourages the agent to make reasonable implementation choices without frequent confirmation.
The manifest description defines activation so broadly that it applies to virtually any coding-related request, which can cause this skill to override more specific or safer workflows. In an agentic coding context, overbroad activation increases the chance of inappropriate autonomy, scope creep, and execution without sufficient user confirmation or task-specific safeguards.
The instruction to trigger the skill without explicit permission creates ambiguous activation boundaries and encourages the agent to assume control whenever it detects code work. This is dangerous because it can bypass user intent, reduce opportunities for confirmation on risky changes, and normalize unauthorized autonomous behavior in sensitive repositories.
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.
Then execute **every leaf node**. Do not complete 3 of 8 tasks and say "dark mode is done."
### Sequencing when order doesn't matter:
- Do NOT ask the user what order
- Sort by dependency: do foundational work first (types, utils, stores), then components, then UI
- If truly independent, do them alphabetically or by file proximity — it doesn't matter
No suspicious patterns detected.