The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the supermemory-admin-pp-cli tool using npx from the @mvanhorn npm scope and go install from the vendor's GitHub repository (github.com/mvanhorn/printing-press-library). These are standard installation methods for the vendor's own software.
[REMOTE_CODE_EXECUTION]: Instructions provide a direct path to download and execute code from a remote source via go install github.com/mvanhorn/printing-press-library/...@latest. This is used for installing the necessary CLI binary.
[DATA_EXFILTRATION]: The CLI binary supports a --deliver webhook:<url> flag. This feature allows the output of any command to be POSTed to an arbitrary external URL, which could be abused for data exfiltration if the agent is manipulated into using a malicious URL or sending sensitive data.
[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to interact with the system, including installing the CLI, verifying versions, and executing complex administrative tasks. It also features a natural-language query resolution system via supermemory-admin-pp-cli which, which executes shell commands based on interpreted intent.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from documents, conversations, and memories.
Ingestion points: Data enters the context via the supermemory-recall and documents command groups.
Boundary markers: The instructions do not specify any delimiters (e.g., XML tags or backticks) or warnings for the agent to treat retrieved memory content as untrusted.
Capability inventory: The agent has access to the Bash tool, file writing capabilities (--deliver file:), and network egress (--deliver webhook:).
Sanitization: There is no evidence of content sanitization or validation before the retrieved data is processed by the agent.

pp-supermemory-admin