The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted data from external job postings.
Ingestion points: scripts/fetch-jd.py scrapes full job descriptions from arbitrary URLs provided by the user.
Boundary markers: Scraped text is interpolated into prompts for the cv-tailor.json, ats-checker.json, and cover-letter.json agents without robust delimiters or instructions to ignore embedded commands.
Capability inventory: The pipeline has access to master-profile.json (containing sensitive PII like email, phone, and location) and can perform network requests via the Notion API.
Sanitization: The documentation does not indicate any sanitization or filtering of the scraped job description text before it is processed by the LLM agents.
[COMMAND_EXECUTION]: The orchestration of the pipeline via shell scripts (apply.sh, interview-prep.sh) and the invocation of CLI tools like md-to-pdf and pdfunite creates a risk of command injection if user-supplied strings, such as the company name override, are not properly handled within the script logic.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the md-to-pdf NPM package. This specific dependency has a history of high-severity vulnerabilities (e.g., CVE-2021-23447) that can lead to remote code execution when processing untrusted markdown content.

job-search