The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script (scripts/extruct-api) to perform API operations. This script is implemented using standard Python libraries and follows secure practices.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to the official vendor domain (api.extruct.ai) for searching and retrieving company data as intended by its design.
[CREDENTIALS_UNSAFE]: The skill handles authentication correctly by using an environment variable (EXTRUCT_API_TOKEN) rather than hardcoding sensitive credentials.
[SAFE]: The instructions and playbooks (e.g., references/finding-companies.md) are focused on legitimate research workflows and do not contain any prompt injection or security bypass attempts.
[PROMPT_INJECTION]: The skill processes data from the Extruct API which may include content sourced from the web.
Ingestion points: Data is retrieved via tables data and deep-search results commands (see SKILL.md).
Boundary markers: Prompt templates in references/column-guide.md use specific structures to contextualize data variables.
Capability inventory: The agent can execute the bundled CLI to perform network operations and update table data within the Extruct platform.
Sanitization: External content is processed through the Extruct API's structured data environment before being presented to the agent.

extruct-api