The Agent Skills Directory

[SAFE]: The skill acts as a governance layer, ensuring research findings are based on a whitelist of high-trust sources (official docs, RFCs, source-repo, and project specs). It does not perform unexpected network operations or handle sensitive credentials.
[COMMAND_EXECUTION]: The skill utilizes Bash to manage file writes to docs/research/ and to append audit logs to .omo/notepads/. It also references a local verification script (scripts/test-citation-discipline.sh) for regression testing, which is standard development behavior.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data retrieved from the web. It implements significant defensive measures, including a mandatory source whitelist, a required citation format, and a strict output schema with YAML frontmatter to prevent the agent from adopting instructions potentially hidden in researched content.
[DATA_EXFILTRATION]: No exfiltration risks were detected. All network-related activity (via the librarian agent) is targeted at well-known, authoritative documentation repositories and registries.

research