The Agent Skills Directory

[SAFE]: The skill's primary functionality is restricted to natural language interaction and local documentation of business decisions. It does not perform sensitive system operations or network requests.
[SAFE]: File operations are limited to reading and writing BUSINESS_CONTEXT.md and saving text-based decision records to a decisions/ directory. These operations are transparently described and serve the intended purpose of the skill.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting data from an external file (BUSINESS_CONTEXT.md).
Ingestion points: Reads BUSINESS_CONTEXT.md in the 'Before Starting' phase to gather company details.
Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore instructions embedded within the context file.
Capability inventory: Limited to writing markdown files to the local directory (context updates and decision logs). No network access, shell execution, or high-privilege capabilities are utilized.
Sanitization: Absent; however, the lack of dangerous downstream capabilities (like code execution or exfiltration) renders this surface low-risk.
[SAFE]: No obfuscation, hardcoded credentials, or persistence mechanisms were detected in the instructions or metadata.

strategic-sparring