The Agent Skills Directory

[SAFE]: The skill uses the mocli utility, which is a tool provided by the skill author, to manage a local database of nicknames. All modification operations (set and remove) are correctly scoped to local configuration and require explicit user confirmation before execution. The skill also incorporates shared security rules from a central configuration file.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-defined remark strings which could theoretically contain instructions intended to influence the agent. Ingestion points: Data is ingested via the output of the mocli remark list command as described in SKILL.md. Boundary markers: The agent is provided with a strict JSON response schema to follow, which helps isolate the data values from the command structure. Capability inventory: The skill is limited to local remark management via the mocli binary. Sanitization: The skill instructions explicitly direct the agent to only use the UID values for subsequent tool calls and not the remark names, which serves as an effective logical filter against injection in the remark labels.

mo-remark