The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts bitwarden_to_profile.py, profile_to_bws_secret.py, and materialize_profile.py utilize the subprocess module to interface with Bitwarden CLI tools (bws and bw). These executions are performed using list-based arguments without a shell, which effectively mitigates command injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill processes profile data from external sources, which is a potential entry point for indirect prompt injection.
Ingestion points: Profile data is loaded from Bitwarden or local YAML files via scripts/bitwarden_to_profile.py and scripts/load_profile.py.
Boundary markers: While the skill normalizes and validates data, it does not define specific prompt delimiters for the final interpolation of profile values into LLM context.
Capability inventory: The skill possesses capabilities for local file writing and executing Bitwarden CLI operations.
Sanitization: Data integrity is maintained by using yaml.safe_load and verifying the data against a JSON schema in scripts/validate_profile.py.

bitwarden-personal-profile