The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the olares-cli binary to perform global searches over local Drive files, Seafile/Sync libraries, and cloud integrations (Google Drive, Dropbox).
[PROMPT_INJECTION]: The skill has a potential attack surface for indirect prompt injection as it ingests untrusted file content during full-text searches.
Ingestion points: Full-text content retrieved from Drive, Sync, Google Drive, and Dropbox via olares-cli search.
Boundary markers: None explicitly defined in the skill instructions; it relies on the agent's interpretation of JSON/CLI output.
Capability inventory: The skill itself is restricted to read-only search subcommands, though it references external skills for file management and settings.
Sanitization: No specific sanitization or filtering of search results is defined within the skill text.
[SAFE]: The skill follows least-privilege principles by maintaining a read-only state for search operations and requiring explicit user confirmation or delegation to olares-settings for index mutations (add, remove, rebuild).

olares-search