The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes "existing findings," which typically include data from external web sources. This creates a potential surface for indirect prompt injection where malicious instructions embedded in the findings could attempt to influence the agent's behavior.
Ingestion points: Reads "Existing findings" as specified in the input section of SKILL.md.
Boundary markers: The skill does not define clear delimiters or specific instructions for the agent to disregard commands within the input data.
Capability inventory: The skill has the capability to write "authorized local artifacts" and utilize "supported contact tools" for sending briefs (SKILL.md).
Sanitization: There is no mention of sanitizing or validating the findings before they are incorporated into the implementation brief.
[SAFE]: The skill references the author's public GitHub repository for documentation and support. This resource is managed by the skill's author and serves as a legitimate informational source.
Evidence: References https://github.com/AgriciDaniel/chatgpt-seo in SKILL.md.

seo-handoff