The Agent Skills Directory

[NO_CODE]: The skill consists entirely of markdown-based instructions for reasoning. It does not include any Python scripts, Node.js code, binaries, or automated installation commands.
[DATA_EXFILTRATION]: No network access or data exfiltration patterns were found. The skill includes explicit security instructions in SKILL.md to ensure that "confidential data stays local" and to "de-identify before anything crosses to an external tool."
[PROMPT_INJECTION]: There are no attempts to bypass safety filters or override system instructions. The skill reinforces agent safety by instructing it to follow "standing rules" and "org instructions" found in the host environment or system prompt.
[INDIRECT_PROMPT_INJECTION]: The skill establishes a process for interpreting external data (user requests and files) to create planning documents. It includes strong mitigations for indirect prompt injection via a mandatory "Confirm, then execute" step, ensuring the user reviews the agent's interpretation before it performs any actions.
[SAFE]: The skill's primary focus is increasing transparency and user oversight in the agent's workflow, which aligns with security best practices for AI agents.

task-brief