The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates a context engine that ingests and recalls untrusted data from external sources, which can influence agent behavior.
Ingestion points: The add_resource tool accepts local file paths, directory paths, public URLs, and Git URLs. The add_skill tool accepts local markdown files or raw content. The memory_recall tool retrieves processed memories from the OpenViking server.
Boundary markers: Absent. The instructions do not specify the use of delimiters, XML tags, or safety instructions to wrap recalled data when it is interpolated into the agent's prompt context.
Capability inventory: The skill possesses the ability to read local files via ov_read, perform network operations to a configured baseUrl, and register new agent skills via add_skill.
Sanitization: Although the skill mentions "sanitized turn text" for capturing history, there are no instructions for sanitizing or validating content retrieved from external URLs or repositories before it is processed or presented to the agent.
[DYNAMIC_EXECUTION]: The add_skill tool allows the agent to import and register new Agent Skills into OpenViking from local files or raw markdown strings. This provides a mechanism for the dynamic extension of agent capabilities at runtime by loading external instructional content.

openviking-context-database