business-brain-setup
Pass
Audited by Gen Agent Trust Hub on Jul 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a structured methodology for knowledge management using Markdown files. It includes a built-in 'Operating Manual' (CLAUDE.md) that establishes rules for data handling, specifically requiring source citation for all claims to prevent the interpolation of hallucinatory or unsourced training data.
- [SAFE]: Command execution is limited to standard git initialization and commits for version control of the created vault, which is appropriate for the skill's stated purpose.
- [SAFE]: The architecture for processing external data from sources like PDFs or web clips includes built-in mitigations for indirect prompt injection risks. It instructs the agent to treat external data as immutable and to explicitly label any information lacking a primary source as an assumption ('supuesto'). * Ingestion points: . /business-brain/_sources/ and . /business-brain/00-inbox/. * Boundary markers: Mandatory YAML frontmatter with sources and type fields. * Capability inventory: Markdown file creation/modification and Git repository management. * Sanitization: Logic in CLAUDE.md instructing the agent to verify all claims against the internal _sources/ directory.
Audit Metadata