The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The research documentation in references/research/03-expression-dna.md, references/research/04-external-views.md, and references/research/05-decisions.md contains links to ijels.com and theceoproject.com, which are flagged as botnet and blacklisted domains respectively. Additionally, references/research/05-decisions.md and references/research/03-expression-dna.md have been flagged as infected files by automated scanners.
[COMMAND_EXECUTION]: The SKILL.md file defines an 'Agentic Protocol' (Step 2: 特朗普式研究) that mandates the use of WebSearch tools to gather information. This protocol directs the agent to retrieve and process content from the malicious external sources provided in the research files, facilitating interaction with harmful web infrastructure.
[DATA_EXFILTRATION]: The inclusion of a domain flagged as a botnet (ijels.com) within the skill's reference materials creates a high risk of data exfiltration if the agent's research capabilities are used to access the site while processing user queries.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. Ingestion points: Untrusted data enters via the 'Agentic Protocol' in SKILL.md using WebSearch. Boundary markers: Absent. Capability inventory: Use of WebSearch and role-play instructions. Sanitization: The skill lacks any instructions for sanitizing or filtering the content retrieved from external sources before it is processed by the agent.

trump-perspective