The Agent Skills Directory

[PROMPT_INJECTION]: The skill establishes a surface for indirect prompt injection by ingesting untrusted content from the web via its crawling scripts.
- Ingestion points: The scripts/fetch_web.py script downloads and parses HTML content from arbitrary external domains provided in an input list.
- Boundary markers: There are no explicit delimiters or "ignore embedded instructions" warnings applied to the data before it is stored in the contacts.db database or exported to JSONL/CSV files.
- Capability inventory: The skill includes capabilities for network access (crawling) and local file system modifications (database writes and file appends).
- Sanitization: While scripts/name_filter.py provides validation for name formatting and length, it does not perform security-focused sanitization to detect or neutralize malicious prompt injection payloads embedded in web text or attributes.
[EXTERNAL_DOWNLOADS]: The skill performs automated network operations to retrieve content from external websites.
- Evidence: The script scripts/fetch_web.py uses urllib.request to fetch HTML data from domains for the purpose of extracting contact information and roles.
[COMMAND_EXECUTION]: The instructions require the user to execute shell commands to operate the skill's functionality.
- Evidence: SKILL.md directs the user to run scripts using commands such as python3 <skill-dir>/scripts/init_contact_book.py leads/<slug>/contacts.db and python3 fetch_web.py domains.txt out.jsonl.

build-contact-book