The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/familysearch.py utilizes subprocess.run to call the macOS security command-line tool. This is used to programmatically retrieve stored API tokens from the system keychain.
[DATA_EXFILTRATION]: The skill accesses sensitive authentication credentials from the environment or the macOS Keychain. These tokens are then transmitted to the external FamilySearch API to authorize genealogical data requests.
[EXTERNAL_DOWNLOADS]: The skill makes authenticated GET and POST requests to api.familysearch.org and ident.familysearch.org. These are the official endpoints for the FamilySearch genealogy service.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the processing of GEDCOM (.ged) files.
Ingestion points: Genealogical data is read from local files using the scripts/gedcom_query.py script.
Boundary markers: The skill does not employ explicit boundary markers or instructions to the agent to disregard instructions that might be embedded within the GEDCOM records during narrative generation.
Capability inventory: The agent has access to network tools and local command execution as part of this skill set.
Sanitization: Extracted genealogical data is not sanitized or filtered for potential instruction-based content before being incorporated into agent responses.

familysearch