The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches and processes untrusted content from the web, which could contain malicious instructions designed to manipulate the agent's behavior. * Ingestion points: Fetches job description content from user-provided URLs and parses professional profiles from LinkedIn during role benchmarking (Phase 1). It also reads markdown files from the local filesystem to build a resume library. * Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the data retrieved from external URLs or local files. * Capability inventory: The skill uses tools for reading files (Glob, Read), writing files, and fetching external web data (WebSearch, WebFetch). * Sanitization: There is no mention of sanitizing or validating the ingested content before it is processed by the model.
[COMMAND_EXECUTION]: The skill uses various built-in tools to interact with the environment as part of its normal workflow. * Evidence: Employs Glob and Read tools for local file access and utilizes WebSearch and WebFetch to gather external information.
[DATA_EXFILTRATION]: The skill has the capability to access sensitive user data (resumes containing PII) and perform network operations. * Evidence: While intended for research, the instructions require the agent to read all markdown files in a user-specified directory and then execute network requests to research the target company and role. This creates a vector where information from the files could be transmitted to external services.

resume-tailoring