The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from roadmap files and research dossiers, creating a potential vector for indirect prompt injection.
Ingestion points: The agent reads roadmap files located at docs/content/docs/roadmap/<slug>.mdx and content from linked research dossiers or the output of the deep-research tool.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to separate external data from its primary operating directives.
Capability inventory: The skill is authorized to write to the local filesystem (updating .mdx files) and perform network operations via git commit and git push commands.
Sanitization: There is no evidence of content sanitization or validation before ingested data is incorporated into the design document and pushed to the repository.

jackin-brainstorm