load-context
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes session context files and project documentation (tasks.md, manifest.yaml) which may contain content from previous interactions.
- Ingestion points: Loads files from the
.context/session/directory and specified project paths during full expansion. - Boundary markers: Employs a specific section mapping and report structure to parse content based on markdown headers, providing a logical separation of data.
- Capability inventory: Limited to reading files, listing directory contents via shell commands, and prompting the user for input.
- Sanitization: Includes logic to handle malformed files by skipping unparseable sections, though it lacks explicit sanitization of the textual content within those sections.
- [SAFE]: The skill performs legitimate session management tasks using authorized tools. The use of internal shell utilities and standard directory paths for session state is consistent with its stated purpose and author context.
Audit Metadata