The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected in the skill instructions.
[NO_CODE]: The skill consists entirely of Markdown and YAML instructions. It does not include or reference any external scripts, binaries, or executable code, which eliminates the risk of malicious logic execution at the code level.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input from fields such as incident_description and people_involved.
Ingestion points: Multiple input fields in SKILL.md are used to populate the prompt.
Boundary markers: Absent; the user data is interpolated directly into the instructional prompt.
Capability inventory: The skill is restricted to text generation only; it does not have access to shell commands, network tools, or file system writes.
Sanitization: No explicit sanitization or input validation was identified.
Conclusion: While the skill lacks boundary markers, the absence of actionable tools or side-effect capabilities means any potential prompt injection is confined to the generated text output and poses no threat to the host system or user data.

restorative-practice-protocol-designer