The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands to perform file management and database operations.
Evidence: The IngestPositions workflow uses mv, rm, and fd to organize files from ~/Downloads/, while SKILL.md uses sqlite3 and uv run python to manage the family_office.db system of record.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external CSV files and local configuration, creating an attack surface for indirect prompt injection.
Ingestion points: Fidelity position and balance CSVs from ~/Downloads/ (processed via head in workflows/IngestPositions.md) and fin-guru/data/user-profile.yaml.
Boundary markers: Absent; there are no specific instructions to ignore malicious content within the processed data.
Capability inventory: The skill has the capability to write to the local database, execute project-specific Python modules, and manipulate the local filesystem (move/delete).
Sanitization: Absent; the workflow relies on string matching within file headers and filenames without validation or escaping.

portfolio-syncing