The Agent Skills Directory

[Unverifiable Dependencies & Remote Code Execution] (MEDIUM): The skill frequently instructs the execution of npx -y kimaki@latest. This command downloads code from the npm registry and executes it without a confirmation prompt. As 'kimaki' is not a verified trusted source, this represents a remote code execution risk, though it is the primary method of using the tool. Evidence: SKILL.md and references/kimaki-quick-reference.md.
[Indirect Prompt Injection] (LOW): The skill processes external Discord messages to control project workflows. 1. Ingestion points: Discord messages and slash commands. 2. Boundary markers: Absent; no instructions are provided to the agent to distinguish between user intent and embedded data instructions. 3. Capability inventory: Shell execution (kimaki send) and file system mapping (kimaki project add) as documented in references/kimaki-quick-reference.md. 4. Sanitization: Absent; no evidence of filtering or validation of external content.

kimaki-expert