The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill instructions focus on providing platform documentation and configuration guidance.
[PROMPT_INJECTION]: The skill uses references/learnings.md to persist and read accumulated platform knowledge, which creates a potential surface for indirect prompt injection. 1. Ingestion points: references/learnings.md, references/platform-guide.md, and references/circleback-api-reference.md. 2. Boundary markers: Absent. 3. Capability inventory: No shell execution or system-level modification capabilities; limited to routing to other skills and reading/writing documentation. 4. Sanitization: Absent. This mechanism is standard for agent memory and does not present a high risk in this context.
[EXTERNAL_DOWNLOADS]: References to external documentation and services are consistent with the skill's purpose and point to legitimate Circleback and CRM provider domains.

sales-circleback