The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents and uses the openclaw CLI and various tools (openclaw openviking status, openclaw config get, openclaw openviking setup) to manage configuration and state.
[DATA_EXFILTRATION]: The add_resource tool is designed to import local files, directories, and URLs into the remote OpenViking server via the /api/v1/resources/temp_upload endpoint. While this is the intended purpose, it involves sending local data to a remote service.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a long-term memory interface, where data retrieved via memory_recall, ov_search, or ov_read is injected back into the agent's context. This presents a surface for indirect prompt injection if the stored data contains malicious instructions.
Ingestion points: External data is ingested through memory_recall, ov_search, ov_read, and ov_archive_search (all defined in SKILL.md).
Boundary markers: The instructions do not specify explicit delimiters or boundary markers for the injected context, though they mention turn sanitization.
Capability inventory: The skill allows file uploads (add_resource), skill registration (add_skill), and reading from the session archives.
Sanitization: The skill mentions that turn text is sanitized before being appended to the OpenViking session, which provides some mitigation against malformed data.

openviking-context-database