saving-workspace-context
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read arbitrary files from the workspace (e.g.,
context/,PROJECT.md,research/) to establish agent context. - Ingestion points: The agent reads content from various workspace directories and files including
context/,.agents/,companies/,docs/,research/, andPROJECT.md. - Boundary markers: Absent. The instructions do not provide delimiters or warnings to the agent to disregard instructions that might be embedded within these workspace files.
- Capability inventory: The skill utilizes file system read and write capabilities, including the ability to create and modify files and generate new agent rules and skills.
- Sanitization: Absent. Data read from external files is incorporated into the agent's reasoning and used to generate new persistent workspace assets without validation or filtering.
- [DYNAMIC_EXECUTION]: The skill instructions direct the agent to dynamically generate and persist new agent skills in
.cursor/skills/or.agents/skills/and new rules in.cursor/rules/based on patterns discovered during conversations. These generated files define behavior for future agent sessions, effectively modifying the agent's instruction set at runtime.
Audit Metadata