The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local commands using zot or cargo run -q -p zot-cli to interact with the Zotero database. These commands are used solely for data retrieval and are constrained by the skill's read-only operating rules.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests untrusted content from Zotero item metadata, abstracts, and notes.
Ingestion points: Data enters the agent context via item get, item children, and item fulltext commands during the research synthesis phase.
Boundary markers: The skill uses templates/report.md and templates/report.html to provide a structured format for the output, which helps separate data from instructions.
Capability inventory: The skill can execute local Zotero retrieval commands and write report files to the local file system.
Sanitization: The skill relies on the agent's synthesis logic to group and summarize data into predefined templates, though it does not specify explicit character-level escaping for ingested paper content.
[DATA_EXFILTRATION]: The skill includes robust protections against data exfiltration by mandating that HTML reports be self-contained and offline, explicitly forbidding CDNs, remote scripts, and network calls such as fetch or XHR.

zot-brainstorm