The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and retrieval of documents using the ingest_documents and query tools, which creates a surface for indirect prompt injection from external data sources.
Ingestion points: The ingest_documents tool referenced in SKILL.md allows the agent to add documents to the retriever.
Boundary markers: The instructions recommend grounding responses in "evidence hits" and preserving identifiers, but do not provide explicit instructions for using delimiters to isolate untrusted content.
Capability inventory: The skill instructions focus on retrieval and ingestion via MCP tools; no shell execution, file system modification, or network capabilities are present in the skill files.
Sanitization: There are no explicit instructions for sanitizing or escaping the content retrieved from documents before it is processed by the model.
[NO_CODE]: The skill consists entirely of markdown instructions and configuration metadata. There are no scripts (e.g., Python, JavaScript, Shell) or binaries included in the distribution.
[SAFE]: The skill is authored by the official vendor (NVIDIA) and points to trusted vendor documentation at docs.nvidia.com. No obfuscation, credential exposure, or persistence mechanisms were detected.

nemo-retriever-mcp