The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes JSON data from audit findings and interpolates it into Markdown reports. This creates a surface where content from external findings files is rendered into documentation.
Ingestion points: scripts/merge-findings.mjs and scripts/render-report.mjs read findings payloads from the local filesystem.
Boundary markers: The reporting templates do not use specific boundary markers or instructions to ignore instructions embedded in the findings data.
Capability inventory: The render-report.mjs script can write generated reports to the filesystem using fs.writeFileSync.
Sanitization: Basic character escaping is performed to maintain Markdown table structure.
[DYNAMIC_EXECUTION]: The scripts/measure-context.mjs utility script uses dynamic import() to load local hook scripts for the purpose of estimating their token usage. This is a maintenance function that operates on files within the project repository.

a11y-core