xberg-extracting-tables
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of external documents (PDFs, images, spreadsheets), which constitutes an inherent attack surface. Maliciously crafted documents could contain instructions intended to influence the agent's behavior during or after extraction.
- Ingestion points: File paths and URIs processed via
xberg extractor libraryextract()calls (documented in SKILL.md and API references). - Boundary markers: Extraction results are returned in structured formats (JSON, Markdown), providing a degree of separation between data and instructions.
- Capability inventory: The tool requires file read and shell execution capabilities to perform document parsing.
- Sanitization: Security limits (e.g.,
max_archive_size,max_content_size) are documented inreferences/configuration.mdto protect against resource-based attacks. - [EXTERNAL_DOWNLOADS]: The skill references the installation of the Xberg toolset and its dependencies from official sources and the vendor's own infrastructure.
- Evidence: References to
pip install xberg,npm install @xberg-io/xberg,cargo install xberg-cli, andbrew install xberg-io/tap/xbergacross multiple reference files. - Context: These downloads target official package registries or the author's verified GitHub organization.
Audit Metadata