The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to execute npx minionsai, which fetches the minionsai package from the public NPM registry at runtime.
[REMOTE_CODE_EXECUTION]: By using npx, the skill triggers the execution of code downloaded from the remote NPM registry onto the local system.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates autonomous agent workflows that ingest and process untrusted data, creating an attack surface for indirect prompt injection.
Ingestion points: The skill processes user-defined task prompts, scheduled routines (e.g., scraping HackerNews), and external web content during autonomous research tasks (e.g., visiting competitor sites).
Boundary markers: There are no explicit delimiters or protective instructions described to prevent the agent from following malicious commands hidden within the data it scrapes or processes.
Capability inventory: The skill environment allows for network access (LLM APIs, web scraping), file system writes in workspace directories (e.g., src/utils/fetch-urls.ts), and local SQLite database interactions.
Sanitization: The instructions do not mention any validation or sanitization steps for external content before it is processed by the agent's logic.
[PERSISTENCE]: The skill provides a 'Routines' feature allowing for the scheduling of recurring agent jobs using cron syntax, which provides a mechanism for maintaining persistent execution of agent-controlled scripts over time.
[CREDENTIALS_UNSAFE]: The documentation provides examples for setting environment variables for OPENAI_API_KEY and ANTHROPIC_API_KEY. These are intended as placeholders for user-supplied configuration and follow safe secret management practices.

minions-hermes-mission-control