The Agent Skills Directory

[SAFE]: The skill employs standard system diagnostic commands such as pkg-config, cat, dpkg, and rpm to inspect the DOCA environment. No unauthorized file access or sensitive data exposure was detected.
[SAFE]: All external URLs point to legitimate NVIDIA and Mellanox documentation and repository sites, which are authoritative sources for the skill's author.
[INDIRECT_PROMPT_INJECTION]: The skill processes output from shell commands, which is a potential surface for indirect injection. However, this is inherent to the diagnostic task, and the skill includes mitigation instructions such as quoting exact output and prohibiting version hallucination.
Ingestion points: Shell command outputs in TASKS.md (e.g., pkg-config, cat, dpkg).
Boundary markers: None explicitly defined for command results.
Capability inventory: Uses shell command execution for diagnostics.
Sanitization: Instructs the agent to use exact quotes and follow a strict anti-hallucination policy.

doca-version