second-brain
Pass
Audited by Gen Agent Trust Hub on May 13, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
sbcommand-line tool to manage capture, querying, and inbox tasks. These operations are restricted to the intended functionality of the personal knowledge management system. - [DATA_EXPOSURE]: The system stores data locally in a SQLite database and configuration files within the user's home directory (
~/.local/share/secondbrain/and~/.config/secondbrain/). This is standard practice for local-first applications and does not involve unauthorized data transmission. - [PROMPT_INJECTION]: The 'Meeting Transcript Processing' feature ingests external text to extract tasks and insights. While this creates a surface for indirect prompt injection (where instructions within a transcript could attempt to influence the agent), the skill instructions focus on structured extraction using specific patterns (e.g., 'TODO:', 'Action:'), which limits the risk.
Audit Metadata