The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses various shell commands such as ls, grep, git, mkdir, test, and zip to interact with the local filesystem and manage the Obsidian vault structure.
[COMMAND_EXECUTION]: In fallback mode, the skill generates a directory slug by piping user-provided $ARGUMENTS through tr and cut. The logic includes character filtering (alphanumeric and hyphens only) which mitigates potential command injection risks.
[EXTERNAL_DOWNLOADS]: The documentation provides installation commands using npx to fetch the skill from the author's GitHub repository (github.com/leoluyi/skills).
[PERSISTENCE_MECHANISMS]: The hooks/README.md file provides instructions for users to manually add a PreToolUse hook to their global ~/.claude/settings.json. This hook executes a local shell script (guard-vault-path.sh) to intercept and validate file write operations.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests data from external sources (WebSearch, WebFetch) and local files (PDFs, images) without using explicit boundary markers or delimiters in the instructions to isolate this untrusted content.
Ingestion points: Web search results, user-supplied $ARGUMENTS, and existing files in the inbox (SKILL.md).
Boundary markers: Absent. The prompt does not use specific delimiters to wrap external data.
Capability inventory: The skill has extensive file system access (ls, grep, Write, Edit, git) and network capabilities (WebSearch).
Sanitization: Content is processed by the agent; filename generation includes character filtering via shell pipes.

learn-loop