The Agent Skills Directory

[COMMAND_EXECUTION]: The skill operates by mapping natural language requests to specific commands for the zot CLI tool. It includes a development fallback to run the tool via cargo run -q -p zot-cli if the binary is not in the system PATH.
[CREDENTIALS_UNSAFE]: While the skill manages Zotero API keys and Library IDs for web-based mutation operations, it includes a strict 'Hard Constraint' that API keys must never be included in outputs, logs, or documentation examples.
[DATA_EXPOSURE]: The skill interacts with the local Zotero SQLite database and local attachments. It maintains security by treating the local database as strictly read-only and routing all modifications through the Zotero Web API or the authenticated local connector.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of Zotero attachments and fetches citation reports from the external Scite service. These operations are within the stated primary purpose of the skill.
[SAFE]: The skill implements an exemplary safety model for 'Security Gates,' categorizing write operations into risk levels (Layer A, B, and C). High-risk operations like merging, deleting, or bulk updates require mandatory 'dry-run' previews and explicit user confirmation before execution.

zot