The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous examples of shell commands using the deepline CLI to perform lead searches, data enrichment, and authentication. These commands are intended to interface with the vendor's platform and external lead generation services.
[EXTERNAL_DOWNLOADS]: The documentation directs users to install the @deepline/cli tool via npm (npm install -g @deepline/cli) and provides a link to the vendor's website (code.deepline.com). This is standard procedure for using the vendor's services.
[DATA_EXFILTRATION]: The skill processes lead data from CSV files and sends queries to external APIs (Apollo, Crustdata, Exa) to retrieve enrichment data. This behavior is consistent with the primary purpose of building and prioritizing prospect lists.
[REMOTE_CODE_EXECUTION]: A Python script is included for scoring leads based on signal data. The script uses standard libraries (csv, json, sys) to process local files and does not perform any dangerous operations or execute untrusted code.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources (job listings via Crustdata and website content via Exa) which is then processed by a Python script. While this represents an attack surface for indirect prompt injection, the current implementation uses simple keyword matching for scoring rather than feeding the raw text back into an LLM for instruction following, which minimizes the risk.

build-tam