The Agent Skills Directory

[SAFE]: No security threats were detected. The skill instructions and associated files are focused on content optimization and business logic without any indicators of malicious intent or dangerous technical operations.
[COMMAND_EXECUTION]: The provided Python script (scripts/conversion_blueprints.py) contains only data structures (dataclasses) and static helper functions. It does not import modules like os, subprocess, or shlex, and lacks any system-level command execution.
[DATA_EXFILTRATION]: There are no network-capable functions (e.g., requests, curl, socket) or attempts to access sensitive file system paths (e.g., .ssh, .env, credentials).
[REMOTE_CODE_EXECUTION]: No external code is downloaded or executed. The skill does not perform package installations or remote script fetching.
[PROMPT_INJECTION]: The instructions in SKILL.md are well-structured for content optimization. There are no attempts to bypass safety filters, extract system prompts, or override agent behavior through injection techniques.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to process untrusted user content (URLs, HTML, or drafts) as specified in SKILL.md, it lacks the capability to execute code or access the network based on that data, mitigating the risk of exploitation.
[DYNAMIC_EXECUTION]: No use of eval(), exec(), or unsafe deserialization was found in the codebase.

geo-conversion-optimizer