The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, untrusted sources including web pages, PDF documents, and source code repositories during the research phase.
Ingestion points: The research methodology described in references/research-method.md utilizes WebFetch, WebSearch, and Read (for PDFs) to gather information from the public internet and user-provided files.
Boundary markers: The instructions do not specify the use of boundary markers or instructions to ignore potential malicious prompts embedded within the gathered research material.
Capability inventory: The skill has the capability to write files to the local directory, call other agent skills, and execute Javascript in a browser context via developer tools.
Sanitization: There is no evidence of content sanitization or validation for the data retrieved from external sources before it is interpreted by the agent.
[COMMAND_EXECUTION]: The skill leverages browser automation tools to execute Javascript for layout auditing purposes.
Evidence: In references/overflow-audit.md, the skill provides Javascript snippets to be executed via chrome-devtools to detect element overflows and manage slide transitions during the QA process.
[EXTERNAL_DOWNLOADS]: The skill documentation recommends that users install third-party agent skills from external GitHub repositories to provide necessary slide generation capabilities.
Evidence: SKILL.md suggests the repository https://github.com/zarazhangrui/frontend-slides as a resource for the user to install if a slide skill is not already present in the environment.

slides-video