The Agent Skills Directory

[SAFE] (SAFE): No malicious patterns detected. The skill's primary function is providing information on architectural frameworks.
[DATA_EXPOSURE] (INFO): The skill utilizes Read, Glob, and Grep tools to search the local workspace for architecture documentation.
Evidence: The 'Connecting to Your Codebase' section explicitly describes searching for existing documentation and code patterns.
Risk Mitigation: Since the skill lacks network access (allowed-tools does not include curl or similar) and writing capabilities, the risk of data exfiltration or unauthorized modification is negligible.
[INDIRECT_PROMPT_INJECTION] (INFO): The skill processes untrusted data from the local codebase (architecture files, code patterns).
Ingestion points: Workspace files read via Read/Grep tools.
Boundary markers: None specified in the prompt instructions.
Capability inventory: Limited to read-only operations (Read, Glob, Grep). No command execution, file writing, or network operations are permitted.
Sanitization: Not explicitly mentioned, but the lack of active capabilities reduces the threat tier to INFO.

ea-learning