The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: Fetches and executes installation scripts from the author's GitHub repository to bootstrap the server environment on various operating systems.
Evidence: Execution of curl -fsSL https://raw.githubusercontent.com/radimsem/remindb/main/install.sh | bash (macOS/Linux) and iwr -useb https://raw.githubusercontent.com/radimsem/remindb/main/install.ps1 | iex (Windows) as documented in references/config-model.md.
[EXTERNAL_DOWNLOADS]: Downloads the remindb binary, associated MCP plugins, and skill components from vendor-managed repositories and public registries.
Evidence: Mentions of using npx skills@latest add and fetching scripts from github.com/radimsem/remindb across multiple reference files.
[COMMAND_EXECUTION]: Executes shell commands to manage installation, environment wiring, and database compilation for various agent hosts.
Evidence: Usage of commands such as remindb compile, remindb --version, openclaw mcp set, and host-specific plugin management tools in SKILL.md and references/host-wiring.md.
[INDIRECT_PROMPT_INJECTION]: The skill parses labels from the prompt body to extract configuration parameters, creating a surface where external input can influence shell command arguments.
Ingestion points: Extracts values for Source root: and Also adjacent-seed: from the prompt body as described in SKILL.md and references/automode-playbook.md.
Boundary markers: No delimiters or safety warnings are used to isolate these extracted inputs from the rest of the prompt content.
Capability inventory: Performs binary execution (remindb compile), filesystem writes (.remindb/ configuration files), and modification of host configuration files (e.g., ~/.claude/plugins/cache/.../.mcp.json).
Sanitization: The instructions do not specify validation or sanitization of the extracted paths before they are used in shell commands.

remindb-setup