The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were found. The skill follows best practices for an AI agent performing research and documentation tasks.
[DATA_EXPOSURE]: The skill instructs the agent to read and write files within specific directories (docs/research and ../raw). This behavior is explicitly defined for managing the research layer and does not target sensitive system files or credentials.
[REMOTE_CODE_EXECUTION]: No remote code execution or unauthorized package installations were detected. The skill mentions 'official-source discovery', which is consistent with its research purpose and does not involve executing untrusted scripts.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes 'raw evidence' from external sources (such as documentation from Google Docs, GitHub, or Notion). However, the instructions emphasize synthesis and classification into a structured 'evidence ledger' rather than executing content from these sources. Standard LLM safety guardrails apply.

research-wiki