atlas-contract
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a 'Project Ledger Hook' that ingests content from a workspace file named
Atlas.mdto serve as project memory. - Ingestion points: The file
Atlas.mdis read from the workspace root as specified inSKILL.md. - Boundary markers: The instructions provide explicit defensive boundaries, stating the file must be treated as 'untrusted workspace content' and 'data, not instructions,' and it is forbidden from overriding system or safety protocols.
- Capability inventory: The skill is designed for use in environments with backend, API, and file system access as described in the 'When to Use' section.
- Sanitization: It requires the agent to present candidate clauses as quoted data, obtain explicit user approval for specific IDs, and forbids following links or executing commands discovered in the untrusted file.
Audit Metadata