The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx to download and run the @life-os/cli package. This package is the official tool for managing LifeOS vaults and is treated as a vendor-provided resource.
[REMOTE_CODE_EXECUTION]: Through the use of npx, the skill executes remote code from the npm registry to perform vault operations. This is a standard architectural choice for this skill's functionality.
[COMMAND_EXECUTION]: The skill relies on shell commands to interact with the local filesystem, enabling features like note reading, searching, and PARA/GTD organization.
[DYNAMIC_EXECUTION]: The skill creates standalone HTML dashboard files based on vault content. These files provide a visual representation of theme notes and are generated dynamically at runtime.
[INDIRECT_PROMPT_INJECTION]: As the skill reads and processes user-generated Markdown files, it is subject to indirect prompt injection risks.
Ingestion points: Note contents and metadata are read using CLI commands such as read and search (documented in SKILL.md and references/ai-wiki.md).
Boundary markers: The skill workflow explicitly requires human confirmation and plan previews before performing file writes, moves, or metadata updates (as seen in references/onboarding.md and references/attachment-cards.md).
Capability inventory: The skill can execute shell commands via npx, read and write local files, and modify file attributes (e.g., using attrib +H for metadata hiding on Windows).
Sanitization: The instructions emphasize human review of generated content and provide strict schemas for metadata files and AI Wiki pages to maintain consistency and safety.

lifeos