eo-project-record
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a
reindexfeature that scans existing Markdown files in thelessons/anddecisions/directories to extract summaries, triggers, and statuses. If these project files contain untrusted content or were modified by an external actor, they could include malicious instructions designed to manipulate the agent's behavior during the data extraction and frontmatter update process. - Ingestion points: The agent reads the body of existing local Markdown files in the
lessons/anddecisions/folders during thereindexandsupersededstatus update operations. - Boundary markers: No specific boundary markers or 'ignore' instructions are defined for the agent when it reads existing file content to generate summaries.
- Capability inventory: The skill has the capability to write to the file system, specifically creating and modifying files within the project root directory.
- Sanitization: There is no evidence of sanitization or validation of the text extracted from the files before it is used to populate metadata and index files.
Audit Metadata