The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in the skill instructions, metadata, or associated scripts. The skill focuses on formatting and prioritizing information to assist users with ADHD.
[PROMPT_INJECTION]: The instructions do not contain bypasses or overrides of safety filters. It includes an 'Invocation boundary' and 'Pre-send gate' to ensure the agent remains within the intended persona and maintains accuracy.
[DATA_EXFILTRATION]: No network operations or commands capable of data exfiltration were found. The skill does not access sensitive system paths (e.g., .ssh, .aws).
[REMOTE_CODE_EXECUTION]: No remote code execution patterns or downloads from untrusted sources were detected. The Python scripts (validate.py, test_skill.py) are used for local package validation and utilize only standard libraries.
[COMMAND_EXECUTION]: The skill mentions authorized implementation in its evaluations (e.g., fixing bugs in local files), but these are test scenarios for the agent's existing capabilities and do not include malicious shell commands or privilege escalation attempts.
[INDIRECT_PROMPT_INJECTION]: The skill defines a formatting layer for user requests. While it processes user input, it does not ingest external, untrusted data from the web or APIs that could trigger secondary injections.

adhd-friendly