The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user text to perform cognitive profiling, creating a surface for indirect prompt injection.
Ingestion points: The skill accepts arbitrary text from the user for analysis (e.g., "profile this writing pattern" or "analyze how this person reasons").
Boundary markers: Absent. The instructions do not specify the use of delimiters (like XML tags or triple backticks) or provide explicit warnings to the agent to ignore instructions embedded within the text being analyzed.
Capability inventory: According to the allowed-tools metadata in SKILL.md, the skill has access to Read and Write tools. This allows the agent to read local files and write the generated profile to the file system.
Sanitization: Absent. There is no instruction to escape, filter, or validate the content of the input text before processing it.
[EXTERNAL_DOWNLOADS]: The skill references external research and author resources.
Fetches technical documentation and research papers from Zenodo (CERN's repository) via DOI links.
Links to the author's official domains, ahkstrategies.net and themindbook.app, for additional tools and research context.

dhdna-profiler