The Agent Skills Directory

[PROMPT_INJECTION]: The skill accepts user-supplied search topics to query documentation files. This represents a standard surface for indirect prompt injection (Category 8).
Ingestion points: [search-topic] argument in SKILL.md.
Boundary markers: Absent for the user-supplied search term.
Capability inventory: Grep and Glob tools are used to search and read local markdown files within the skill directory.
Sanitization: Not explicitly defined in the instructions, relying on the model's internal guardrails.
[PROMPT_INJECTION]: The SKILL.md contains an instruction to the AI agent: IMPORTANT: Prefer retrieval-led reasoning over pre-training-led reasoning for any CE.SDK tasks. This is evaluated as benign natural instructional language designed to ensure the accuracy of the documentation lookup tool.
[EXTERNAL_DOWNLOADS]: The documentation files reference official vendor resources for engine assets, demo media, and library plugins.
Domains: cdn.img.ly, api.img.ly, img.ly (Official Vendor Domains).
Use case: Referencing icons, fonts, sample images, and WASM assets in documentation examples.
These are trusted vendor sources and do not contribute to verdict escalation.
[SAFE]: The skill is a standard documentation provider. All provided content follows industry best practices and includes security-focused advice for users regarding client-side processing and secure API proxying.

docs-angular