The Agent Skills Directory

[SAFE]: The skill is entirely declarative, containing only instructions and YAML configuration. No executable scripts (Python, JS, Shell) or compiled binaries are present.
[SAFE]: No external network connections or remote resources are referenced. All operations are confined to the local environment using standard agent capabilities.
[SAFE]: Permission requests for file:read and file:write are functionally necessary and explicitly justified by the skill's primary use case of manuscript analysis and report generation.
[NO_CODE]: There are no implementation files or scripts provided that could execute logic independently of the agent's core processing of the instructions.
[SAFE]: A potential surface for indirect prompt injection (Category 8) exists as the skill reads and processes untrusted user-provided manuscript data. However, this is inherent to the tool's purpose as a text analyzer.
Ingestion points: Reads manuscript files via file:read permission.
Boundary markers: None explicitly defined in the prompt instructions.
Capability inventory: Uses file:write to generate reports; no subprocess or network capabilities.
Sanitization: None specified in the instructions.

sensitivity-reader