The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch version metadata from https://skills.opper.ai/versions.json and potentially an updated skill definition from https://skills.opper.ai/opper-mcp/SKILL.md. These resources are hosted on the vendor's official infrastructure and represent a standard update mechanism.
[DYNAMIC_EXECUTION]: A version-check routine is defined where the agent compares local version tags with a remote JSON file. If they differ, the agent is directed to load and follow instructions from a remote SKILL.md file. This constitutes dynamic loading of instructions from a vendor-controlled remote source.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes data from external sources and performs sensitive operations.
Ingestion points: Fetches remote skill instructions from skills.opper.ai; processes inference responses via call_model; reads platform documentation via get_guide; and retrieves execution traces via get_trace.
Boundary markers: No specific delimiters or "ignore previous instruction" markers are defined for the untrusted data returned by the tools.
Capability inventory: The skill can execute shell commands via opper-cli, create API keys, manage control plane rules, and perform billed model inference.
Sanitization: No explicit sanitization, validation, or escaping of external content is specified before the data is interpolated into the agent's context.

opper-mcp