The Agent Skills Directory

[SAFE]: The analyzed skill is composed entirely of informational Markdown content.
No executable scripts (.sh, .py, .js) or binaries are included in the skill package.
The instructions focus on business logic, financial risk assessment, and regulatory compliance.
[NO_CODE]: The skill serves as a template and procedural guide for risk management.
It does not define any new tools or provide scripts for the agent to execute.
All tasks are intended to be performed using the agent's native capabilities for reading and writing text within a business plan context.
[INDIRECT_PROMPT_INJECTION]: The skill outlines a workflow for ingesting external business documents (contracts, logs, records).
Ingestion points: SLA definitions, monitoring evidence, and dispute records (SKILL.md).
Boundary markers: Explicitly defines 'Capability and Permission Boundaries' limiting the agent to read-only analysis.
Capability inventory: No dangerous tools or subprocess calls are requested or provided by the skill.
Sanitization: Not applicable as no automated processing logic is defined.

saas-agent-sla-risk