The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from externally generated research specifications (e.g., research_spec_*.md) and interpolates this content into grant proposal sections. This constitutes an indirect prompt injection surface where the source data could potentially influence agent behavior if it contains instructional text.
Ingestion points: The skill reads input from the file path provided in the --input argument or automatically locates files in quality_reports/specs/ during Phase 0.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to wrap the external content during the scaffolding process.
Capability inventory: The skill is granted Write and Task capabilities, which are used to generate markdown files and invoke secondary skills like /data-management-plan and /capture-environment.
Sanitization: The skill does not explicitly sanitize or filter the content extracted from the research specs, although it does include a directive to not 'fabricate science' if the spec is missing.

grant-proposal