The Agent Skills Directory

[SAFE]: The skill is designed as a persona-based assistant for business consulting. It operates within a defined local directory (~/Clawic/data/consultant/) and includes explicit instructions to maintain factual accuracy and avoid storing sensitive information such as credentials or secrets.
[DATA_EXPOSURE]: Technical controls are present in the form of instructions (memory-template.md) that forbid the storage of private data not required for work. The skill does not attempt to access environment variables, shell history, or other sensitive user configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface through its discovery process where it collects client context.
Ingestion points: Discovery interviews and context updates logged in memory.md.
Boundary markers: Structure is provided by templates in deliverables.md, though explicit delimiters for raw user input are not mandated.
Capability inventory: File system writes restricted to the skill's own workspace folder.
Sanitization: The skill relies on quality control checklists (quality-gates.md) to ensure the agent verifies logic and evidence before outputting recommendations.

Consultant