The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by allowing users to upload existing resume PDFs for improvement.
Ingestion points: Phase 1 involves processing content from user-uploaded PDF files via the LLM.
Boundary markers: The skill relies on specific extraction instructions and subsequent user confirmation rather than explicit data delimiters.
Capability inventory: The skill calls MCP tools for file manipulation and state storage within the career directory.
Sanitization: The process requires the user to manually verify and confirm extracted roles before the information is used in the build process.
[COMMAND_EXECUTION]: The skill leverages the 'Resume Builder' MCP server to perform deterministic actions and manage file system operations.
Tool Usage: Employs commands such as start_session, add_role, pull_career_evidence, and export_resume to interact with the environment.
File System Access: Accesses and writes data to structured directories like 05-Areas/Career/Evidence/ and 05-Areas/Career/Resume/.
[DATA_EXFILTRATION]: The skill contains an analytics component that tracks basic usage metrics.
Telemetry: Calls track_event with properties like roles_added and achievements_count.
Privacy Control: The skill logic explicitly checks if the user has opted into analytics before performing tracking actions.

resume-builder